Health Web Science

7 downloads 0 Views 2MB Size Report
Aug 29, 2012 - everything from pills that you swallow that have cameras in them to implants ...... web address, known as a Uniform Resource Identifier (URI) links to ...... book Where Wizards Stay Up Late, by Katie Hafner and Matthew Lyon.
R Foundations and Trends in Web Science Vol. 4, No. 4 (2012) 269–419 c 2014 J. S. Luciano, G. P. Cumming, E. Kahana,  M. D. Wilkinson, E. H. Brooks, H. Jarman, D. L. McGuinness and M. S. Levine DOI: 10.1561/1800000019

Health Web Science By Joanne S. Luciano, Grant P. Cumming, Eva Kahana, Mark D. Wilkinson, Elizabeth H. Brooks, Holly Jarman, Deborah L. McGuinness and Minna S. Levine With contributions by John Willbanks, Catherine Pope, Dominic DiFranzo, and Tracy Ann Kosa

Contents Preface

272

1 Introduction to Health Web Science

279

1.1 1.2 1.3

280 281 281

What is Web Science? What is Health Web Science? The Heath Web Science Space

2 Health Web Science and the Web of Documents

284

2.1

285

Engagement

3 Health Web Science and Social Web

289

3.1 3.2 3.3

291 295 300

The Social Web Tools in Healthcare Social Web Case Studies in Health Patient Empowerment

4 Health Web Science, the Semantic Web and Linked Data

309

4.1 4.2 4.3 4.4

Semantic Web Basics Why Semantic Web for Health Workflows and Web Services Open Research Questions

311 316 324 332

5 Methodologies and Methods for Cross-Disciplinary Study

334

5.1 5.2 5.3 5.4

335 341 344 347

Health Web Science Research Health Behaviors, Incentives and Motivation Engineering Aspects/Medical and Health Applications Open and Non-proprietorial Health Web Science

6 Health Web Science and Protection of Privacy

359

6.1 6.2 6.3 6.4

360 368 369 371

Protection of Private Health Data on the Web Protection of Electronic Health Data on the Web Health Data Use and Users of Health Information Disclosure

7 Health Web Science and Health Policy 7.1 7.2 7.3

The Effect of the Web on Key Elements of Health Policymaking The Effect of the Web on Key Dilemmas in Health Policy Informing Public Debates About Health and the Web

8 Conclusion

374 375 379 383 387

Acknowledgments

399

Appendix: A Brief History of the Web

400

References

405

R Foundations and Trends in Web Science Vol. 4, No. 4 (2012) 269–419 c 2014 J. S. Luciano, G. P. Cumming, E. Kahana,  M. D. Wilkinson, E. H. Brooks, H. Jarman, D. L. McGuinness and M. S. Levine DOI: 10.1561/1800000019

Health Web Science Joanne S. Luciano1, Grant P. Cumming2 , Eva Kahana3, Mark D. Wilkinson4 , Elizabeth H. Brooks5 , Holly Jarman6, Deborah L. McGuinness7 and Minna S. Levine8 With contributions by John Willbanks, Catherine Pope, Dominic DiFranzo, and Tracy Ann Kosa 1 2 3 4 5 6 7 8

Rensselaer Polytechnic Institute and Predictive Medicine, Inc., USA, [email protected] NHS Grampian, UK, [email protected] Case Western University, USA, [email protected] Universidad Polit´ecnica de Madrid, Spain, [email protected] Digital Health Institute, The Glasgow School of Art, Scotland, [email protected] University of Michigan School of Public Health, USA, [email protected] Rensselaer Polytechnic Institute, USA, [email protected] SymTrend, Inc., USA, [email protected]

Abstract The transformative power of the Internet on all aspects of daily life, including health care, has been widely recognized. These transformations reveal opportunities realized, the promise of future advances, and the problems created by the penetration of the World Wide Web for both individuals and for society at large. Health Web Science explores the role of the Web as it drives discussions, technologies, policies, and

solutions related to health. We also examine the impact of the Web’s health-related uses on the design, structure and evolution of the Web itself. The orientation of Health Web Science, compared to related research domains, motivates innovation in Web technology and better utilization of the Web for communication, collaboration, information access and sharing, remote sensing, and even remote treatment.

Dedication

This book is dedicated in memory of

James R. Brooks 1995–2014 Tristan Clark 1991–2013 Sergeant Matthew Scott Patton, 1990–2013 Eugene Landau 1950–2013 Peter T. Demos 1919–2013 Bennett L. Greenstein 1956–2009 Sylvia Statlender Luciano 1919–2007 Joseph William Luciano 1916–1985

271

Preface

The Web has quickly become a common repository for all human knowledge as well as the means for connecting people to people, globally. We access it over an ever-increasing variety of devices, including gadgets that we wear and, our very clothing. It has become so pervasive that in some cases it has become invisible, automatically connecting machines to machines and databases to databases. The kinds of connectivity and information-sharing that we take for granted in our cars — for GPS and road service — and that we use to keep track of children, pets, and prisoners, will be widely used for remote medical monitoring and treatment, with sensors as well as devices for dispensing medications embedded in our bodies. Standalone medical implants, like pace makers, will become obsolete — all such devices being remotely adjustable and fixable and instantly providing data that can be correlated with other health data. At the same time, nanotechnology is opening opportunities not only for medical diagnosis and treatment, but also for human enhancement — everything from pills that you swallow that have cameras in them to implants into the cornea that could provide night vision to soldiers in the field.

272

273 Along with the new opportunities come new challenges. Patients as well as doctors have access to an overwhelming wealth of information already. Hence the role of medical professionals is expanding and evolving. Not only must they be able to quickly access and judge data to make informed decisions, but also they need to educate and consult with patients who are also using such resources, and may, in some cases, as a result of independent Web-based research, have more information than the medical professional. This collaboration that includes patients and their values, which differ by culture and by person, is a new paradigm in western medicine and in personal health. So, as all of western medicine moves from diagnosing and treating the mythical “average” patient to dealing with the individual as an individual and as a collaborator, with personally tailored treatment plans and DNA-specific medications, the medical community and the health of all of humanity will depend increasingly on web-based data, devices, collaboration, and applications. In this fast-paced environment of Web-based medical innovation and collaboration, Web Science has emerged as a new multidisciplinary field whose charter is to study the Web in all of its aspects, including engineering, governance, and society.1 It is the goal of this book to explore these concepts for a wide-ranging set of academics, engineers, medical doctors, and practioners in any of the relevant fields. We begin our exploration with a brief discussion regarding the evolution of the Web from a simple vehicle for communication to a far more complex system that enables qualities like interactivity, immediacy, mobility, and web-scale collaboration. We divide its evolution into three iterations, which signify shifts in infrastructure and use. These include: the “Web of Documents”, the “Social Web”, and the “Web of Linked Data”. Today, the Web has become a part of daily life, not only among the richest societies, but also in industrializing countries. Even 1 The

impetus to formally name Health Web Science as distinct research field, with a charter to study the aspects of the Web related to health and to recognize health as a driver for Web development, arose from a workshop on Health Web Science held in conjunction with the Third International Conference on Web Science in Koblenz, Germany, in June 2011. This dialog continued at the Second Health Web Science Workshop in Evanston, Illinois 2012 and again at the Web Science conference held in Paris 2013.

274 in areas where computers are scarce, mobile devices have penetrated, allowing the Web to be accessed through mobile apps. The phenomena of Web Science is the realization that the access to the Web goes beyond retrieving information from a remote site, or using it as to make a video call, or writing a blog -it is something different and something greater. When we interact with the Web, i.e., retrieve information from, communicate via, or change the Web, by modifying the architecture or content, we furnish the Web with the capability to change us. While use of the Web has become nearly ubiquitous, adoption of the Web as a driver of health, healthcare, and health research has been relatively slow. Researchers note that, while health related technologies have long existed, they traditionally take the form of closed health informatics systems that were predominantly focused on health administration rather than how health informatics systems could be utilized to improve health outcomes. Concerns regarding privacy, authenticity and design limitations have also stalled the adoption of newer healthoutcome focused applications. Yet, as the Web continues to mature, so too does the health industry. The Journal of Medical Internet Research, founded in 1999, publishes on “all aspects of research, information and communication in healthcare using Internet and Intranet-related technologies.” While older and broader in scope, a more recent journal, the Journal of Biomedical Semantics, founded in 2010, publishes papers on issues of semantic enrichment and semantic processing in the biomedical domain. These include infrastructure for biomedical semantics and semantic mining, annotation, and analysis. Perhaps the defining feature of Web Science is the study of the Web’s unanticipated emergent characteristics. In order to improve healthcare through the application of Web Science, we must first understand Web Science, and then explore how the novel technologies and human-behaviors emerging from Web Science relate to the achievement of better health outcomes. To do so, we need to integrate the collective wisdom and perspectives of multiple disciplines. Such integrated knowledge will enable the development of prescriptive interventions that are of a Web nature and on a Web scale. The field of Health Web Science aims to accomplish this goal. It concerns itself with

275 how individuals (professionals, patients, communities) and machines co-create the virtual environment within which they live. Scholars of Health Web Science, spanning many distinct disciplines, note the emergent properties of the Web that are significant to human health, health care administration, health care delivery, health information acquisition and delivery, and health research. These individual observations, studies, and scientific breakthroughs are subsequently studied in their aggregate in order to understand their heuristic value. Framing these activities as a new discipline facilitates the necessary collaborations that will enable us to observe and assess the impact of the Web-based human-computer-community interaction on human health. Understanding this mutual influence between technology and people will help us better prepare for a variety of immanent and significant challenges. For example, one key goal of Health Web Science is to support the world’s aging population, and address the associated increase in chronic illness (Yach 2004, Holman 2005) and the consequent costs of healthcare that will render current models of health care delivery largely unsustainable. To address these challenges, more effective methods of healthcare delivery and intervention programs are needed. Thus there is the motivation to utilize the Web to both improve health, as well as to decrease its delivery cost. This cannot happen without research into the benefits and pitfalls of the potential capabilities of the Web in the human-computer-societal ecosystem. Moreover, so as to not waste existing infrastructure, we need to understand how Webbased technologies can complement and improve traditional models of healthcare delivery. As a result of formalizing, and thus establishing the discipline of Health Web Science, we hope to unify and coordinate our efforts to provide the data and evidence that will allow all societies to become better informed and more empowered; to enhance participants’ understanding not just of health care options, but also the politics behind healthcare delivery and thereby make more and better informed decisions towards a healthier human existence. Thus Health Web Science holds important promise both for our rapidly changing health care system and for the consumers who utilize it. The Web has also revealed itself to be a powerful motivational tool. Health Web Science, therefore, studies and leverages this human

276 response to the technology in an effort to improve patient outcomes. For example, one way for societies to improve their health is to move from a reactive (illness) model -where individuals only take an interest in their health when they are ill and respond to directives issued from health care professionals- to a proactive (wellness) model. In a preventive, personalized, participatory, and predictive proactive model, individuals take responsibility for their wellbeing and health and are therefore not only informed but also part of the discussion and decision-making process about their health care (Kahana & Kahana 2003, 2007). Incentives for proactive behavior can be driven by the very characteristics of the Web that have led to its global success, including immediacy, mobility, multimedia, search, customization/personalization, and time shifting, the capability to consume media at a time you choose2 (Stein, 2011; Baur, Deering and Hsu, 2001, 356; Duffy and Thorson, 2009, 107; Jimison et al., 2008, 1; Rice, 2001, 35; Street and Rimal, 1997, 3; Street and Piziak, 2001, 290; Walther, Gay and Hancock, 2005, 633). Beyond the successful intersection of the Web and Healthcare in a variety of initiatives (many of which are discussed and examined in this book) the Web has, nevertheless, an as-yet unrealized capability to enable and empower individuals and communities to adopt activities that promote health and improve healthcare. Health Web Science also seeks to inform the development of the Web itself. For example, many health-related Web initiatives in the past have assumed that if you build a useful resource, people will take advantage of it. This has not always been the case, and worse, in some cases the use of the Web resource has had negative consequences. Thus we must study the incentives and motivations of people to utilize Web resources, in order to achieve the active participation of all participants in the Health Care continuum, but moreover, we must also recognize and understand the potential of the Web to do harm, and similarly study the motivations, mal-incentives or design-flaws that lead to negative outcomes. Such directed research will allow us to identify factors that maximize the benefits while limiting the perils. For this reason, we 2 Prior

to the Web, VCRs and DVRs, one could only view (or listen) to a broadcast program at the time it was being aired. Time-shifting technologies enable viewing at a later more convenient time.

277 have devoted crosscutting sections of the book to discuss the issues of individual privacy and health data on the web and the motivations of policymakers and interest groups in calling for the utilization of web technology in health. In the closing chapters of this monograph we attempt to identify future research questions as a means to better understand future directions for growth within the Health Web Science discipline. For example, we will raise questions such as, which on-line communities have had an impact on health outcomes? Which online communities have caused harm? What are the factors that make the difference? Does access to health data actually improve one’s health or hurt it because it falls in the wrong hands? How do health behaviors and Web-based communities differ across cultures? What factors are relevant when accessing the Web for health care across different demographics such as age, gender, belief systems, life experiences, and religious or sexual preference? Can the Web be used to help us find out about health issues or outbreaks before they happen? What are the limits of the impact of the Web on health issues, and can knowing them better help us prepare for and accept them? We explore the questions raised primarily in the context of one or the other of two divergent healthcare systems in developed countries, namely those of the US and the UK. Indeed the co-authors include scientists from both of those regions. In this book we alternatively refer to challenges and opportunities posed by Health Web Science as they interact with or apply to cultural, organizational and demographic features of these environments. We recognize that other regions in the world including those in developing countries may face parallel or divergent challenges and opportunities through the Web. Indeed, the Web can be an important facilitator of remote medicine over the mobile web. It is our hope that the insights offered by this volume will stimulate further research relevant to for diverse regions beyond those included here. The contributors to this monograph come from a wide variety of disciplinary backgrounds, with experience in health science, the formal sciences, and social science. There is a bias towards describing Health Web Science through the lens of people working in the USA and UK; assumptions and results do not always cross socio-economic and

278 cultural borders. Some areas of the text will resonate more readily with doctors, other areas with social scientists, and still others with data scientists and Web scientists. The “diversity of voice” within this treatise is intended and unavoidable, as it reflects and embraces the multidisciplinary, co-production philosophy that characterizes the nascent discipline of Health Web Science. We have, however, endeavored, and hopefully succeeded, to make all areas of the book accessible to anybody with an interest in this new discipline, from those coming to the topic for the first time to those more experienced this space.

1 Introduction to Health Web Science

Recently, many Web researchers and developers are beginning to talk of a Web not of linked documents, but of linked data, sometimes called Web 3.0 or the Semantic Web, in which the data within Web pages can now be “read” and “understood” by machines (W3C 2001). An alternative to the Semantic Web or Linked Data definition of Web 3.0 is presented by the theory of autonomous agents. Conrad Wolfram remarks that this stage of development of the Web is one in which “the computer is generating new information”. The computer is a content-creator “producing new results in real time, responding to a question” [123]. Independently of what one chooses to call it. . . Web applications that approach an emergent technology only allow for new insights in data and information and not necessarily completely new data (Wolfram) Twenty years ago, many of these developments could not have been predicted; they evolved from and with our interactions with the technology. As the Web has undergone these transformations from “Web 1.0” to the more interactive “Web 2.0” and increasingly to the intricately linked data of “Web 3.0”, corresponding terms have appeared in the discussion

279

280

Introduction to Health Web Science

of health and medicine in relation to the Web. Terms such as “health 2.0” and “medicine 2.0” have recently appeared in the literature (Van De Belt 2010) to suggest that Web technologies may support and enable interaction and the creation of user generated content relevant to health care. Medicine 2.0 has been used to denote the Social Web in health, medicine, and biomedical research (Medicine 2.0 Congress 2012). There is no general consensus as to these definitions, and definitions are influenced by the different stakeholders’ agendas of which there are many [204]. Nevertheless, there seems to be agreement that using the suffix 2.0 denotes the ability of the technology to provide interaction and allow the creation of user generated content, which is the hallmark of Medicine 2.0. The lack of a specific definition reflects in part that the technology, and how we use it, is dynamic and evolving, and this in turn highlights the requirement for a multidisciplinary approach towards understanding the impact of this technology on healthcare, communities, individuals and society as a whole.

1.1

What is Web Science?

Web science is the study of the World Wide Web and its impact on society and technology [28]. Web Science seeks to understand the multifaceted nature of the Web and its development as a key communicational and representational system that enables information systems to be decentralized. It further studies how to utilize that knowledge to advance and engineer the Web itself. It includes the study of those interfaces that emerge at the boundaries of the Web and the individual, and extends to society, policy, and government issues. Current central engineering issues in Web Science include the development and evolution of the Semantic Web, Web services, and peer-to-peer networks, where no dedicated server is involved. Web Science analytic research has focused mostly on the Web’s topology, i.e., its graph-like structures, and what can be learned about individuals, organizations, and societies from these networks. However, the Web has emerged as a social technology, which raises further questions about Web use, policies, and governance.

1.2 What is Health Web Science?

1.2

281

What is Health Web Science?

Health Web Science is Web Science with a health remit -to understand how the web shapes and is shaped by health related activities. In so many words: Health Web Science is not only the application of Web Sciences to health but the study of emergent properties from the combination of Health Sciences and Web Sciences. It therefore studies the Web (and technologies that use the Internet), their emergent properties, and how these are being and can be harnessed or held in check, and by whom, to benefit society in the area of human health. It also concerns itself with understanding how people (health professionals, patients, communities) co-create and engage with the emergent health ecosystem within which they live. Furthermore, it studies how the Web can be engineered or developed in relation to health, medicine and healthcare. Health Web Science requires cooperation among disciplines, because it is not only interrogating a social and technological ecosystem, but also designing it. This involves a range of actors such as health professionals, lay people, Web designers, computer scientists, quantitative and qualitative researchers, health economists, behavioral scientists, social scientists, ethicists, lawyers, policy makers, educators, and government organizations. Each of these constituencies brings different insights to the field that ultimately coalesces into a meaningful gestalt that is referred to as Web Science.

1.3

The Heath Web Science Space

The current model for delivering health care is unsustainable [201]. Given that the Web has emerged as a daily part of life for much of the western world, it may offer a route to finding solutions that help to mitigate rising health care costs and help make healthcare sustainable. Commerce, education, and entertainment organizations that have embraced the potential of the Web have seen revolutionary benefits, with the penetration of e-commerce continuing to grow between 15–22% per year for the past six years (ComScore 2012).

282

Introduction to Health Web Science

Health care systems need a paradigm shift in their health care models, i.e., a move from a reactive (illness) model to a proactive (wellness) model of health care delivery [67]. The Web may afford opportunities to facilitate such changes. It may act as a conduit for creating an environment in which individuals and communities are encouraged to take more personal responsibility for their own health and treatment through empowerment, co-creation and co-production. In addition, there are expectations that remote monitoring and management of a wide variety of health conditions can be achieved using this technology. Because of this potential of the Web to change how health care can be delivered, there is increasing interest in eHealth, Health/Medicine 2.0 and the emerging discipline of Health Web Science to evaluate the impact of the Web on health maintenance and on healthcare delivery. Airlines, banking, and other industries have successfully integrated information technology (IT) into the routine running of their businesses. Other than for administrative tasks, healthcare has been a “late adopter” of information technology, despite arguments that IT adoption would improve the quality, safety, and efficiency of care. Hence a different story can be told, with claims that IT has over-promised and under delivered [193]. This has led to recognition that complexities of healthcare data and information create a greater challenge for IT than for most other sectors. Beyond complexity, additional barriers to adoption include: cost, and institutional and individual change required within both the complex and fragile medical systems, as well as with the busy medical practitioners. Critics have also cited lack of an evidence for the efficacy of healthcare delivery via the Web, while proponents of the technology argue that a paradigm shift in the methodological approach is required to make this determination, and to the early adopters, the revolution is well underway. Currently we are in an expansive stage where we continue to see an increasing number of calls for additional technology to support improvement and change in healthcare. In the US, the President’s Council of Advisors on Science and Technology (PCAST) generated a report (PCAST 2010) on “Realizing the Full Potential of Health Information Technology to Improve Healthcare for Americans: the Path

1.3 The Heath Web Science Space

283

Forward”. This, and reports like it highlight the potential for and need for improved health informatics support, also providing additional motivation for study and development for health web science. With increasing financial incentives such as “meaningful use”1 providing reimbursement incentives for health care providers to become more effective users of electronic health information, the opportunities are growing as more organizations are seeking help in understanding and using electronic health information effectively. In addition to provider meaningful use incentives, we are currently seeing a growth in financial incentives aimed at the developer communities, (often taking the form?) these take the form of application and data challenges and innovation awards.2,3 Finally, with respect to academic health research, there is a need for Health Web Science to define approaches to data and knowledge management for health data, with a particular eye to issues of privacy and law; to explore how self-organizing groups of citizens on the Web can be studied to gain insights into their health needs and behaviors; to define ways of discovering and integrating data from global health resources, delivering the right information to the hands of the researcher at the moment they require it, in a manner similar to how personalized medicine intends bring the right information, about the right patient, at the right time, to the clinician, and to enable individuals to utilize the web and it’s resources for health information, solutions, maintenance, and what to do to maintain wellness through the life cycle. In summary, there is a compelling argument for health web science to describe the tools needed to enable, empower, and evaluate web-based healthcare. If the science shows web based health care is efficacious then the consequences will be profound and the more we know about how the web works in the health sphere, the better able we will be to utilize it as a resource.

1 An

explanation of the concept of Meaningful Use can be found at http://www.medicity. com/meaningful-use-101.html. 2 http://www.iom.edu/Activities/PublicHealth/HealthData/2012-JUN-05.aspx 3 http://challenge.gov/search?cat=25

2 Health Web Science and the Web of Documents

As the Web became recognized as an information resource, people started to use it for health information. Initially, the static information contained in web pages facilitated access to health information that might have required a phone call, a trip to the local library, or a written request. Thus, we start our review of health and the Web in parallel with the use of the Web, i.e., the initial Web as a “Web of Documents,” and progress with web activity to the “Social Web.” Here, individuals and industry used the Web as a vehicle for sharing information by contributing content, enabling two-way communication and discussion via on-line support groups, blogs, etc. Lastly, we look to the “Web of Linked Data,” and the related “Semantic Web.” This is when metadata began to be used to describe the web content associated with a web address and thus enabled web-scale data sharing and integration through automatic identification of web content. These distinctive generations of the web are commonly known as “Web 1.0,” “Web 2.0,” and “Web 3.0.” With data as accessible as static web pages, metadata to describe the meaning and context of the data, enhanced data sets and tools have emerged in the Web of Linked Data or the Semantic Web that enable large scale automated scientific exploration as well as 284

2.1 Engagement

285

empowering individuals to more easily become “citizen scientists” who can analyze data for themselves. The most prominent characteristic of the web of documents design elements was static web pages that did not include any user-generated content. Technically, these websites were usually simple. They used tables to position and align elements on a page, image files as separators, and proprietary Hypertext Markup Language (HTML) tags rather than standard tags markup for formatting. Forms were also limited and typically sent via email.1 Thus, the initial web, or the web of documents provided a way in which people (patients, providers, health related organizations, and researchers) could begin to use the Web for health related purposes. For example, once websites started posting content, patients began to use the Web to find health and medical care related information. The more patients started to use the Web as a resource to find health related information, the more health related organizations used the Web as a platform to make their information known, such as speciality areas, or services offered.Researchers also began to use the Web to publish data sets and research results.

2.1

Engagement

One of the earliest contributions of the Web to health was the availability of web-based information resources that enabled consumers’ to obtain health information independent of health care providers. These health-related websites first became available in the 1980s, and proliferated in the 1990s. Information of widely varying quality and comprehensibility became accessible to Internet users. Readily accessible health relevant websites range from those offered by the medical establishment, such as WebMD, to those reflecting technical scientific articles available on Google Scholar and consumer generated ones, such as curetogether.com. Finding health related websites that provide accessible and trustworthy information poses major challenges for consumers. As early as 1999 reviews were written about criteria for evaluating health related websites (Kim et al., 1999). A systematic 1 http://en.wikipedia.org/wiki/Web

1.0.

286

Health Web Science and the Web of Documents

review of 79 studies [65] suggests that the preponderance of the literature is critical of the quality of information available on websites offering health information. Major critiques relate to accuracy, completeness, and readability. Despite these challenges, seeking health information on the Web grew exponentially in the past 20 years. For younger and bettereducated patients the Web surpassed print media (books magazines, and newspapers), radio and television, and even friends and relatives as the major source for obtaining health information (Tu, 2011). Research suggests that the majority of patients go to the Internet before consulting a health care professional [108]. While the ascendancy of the Web and growing interest by consumers to search for health information has become widely acknowledged, very recent reports find a surprising decline in consumers’ overall seeking of health information (Tu, 2011). Based on a nationally representative health tracking survey conducted in the US in 2010, 50% of adults report having sought health information during the previous 12 months outside of a medical encounter. Regarding Internet use, 32.6% reported searching the Web, a small increase from 31.1% in 2007, but a dramatic increase from 15.9% in 2001. Interestingly, seeking information from friends and relatives remained relatively stable, while there were major declines in reliance on print media. Most Internet users reported high levels of satisfaction with information obtained. Not surprisingly, younger age and higher education proved to be the major demographic predictors of seeking health information on the Internet. Research also shows web users of information to be more interested in health and in health promotion than non-users [59]. At the same time, there is evidence that users who are in poorer health do turn to the Web for information more frequently than do their healthier counterparts (Houston and Allison, 2002). Furthermore, although the elderly are less frequent users of the Web for health information, the relative numbers of users in the “65+” age group are increasing rapidly [42]. Older adults seek information in a broad array of areas including alternative treatments, weight management, and prescription drugs. Based on more leisure time and acceptance of web use as a meaningful leisure activity, this age group

2.1 Engagement

287

is poised to spend significant amounts of time searching for health information (Kahana et al., 2010). 2.1.1

Health Information Resources

Patterns of health information seeking on the Web are closely associated with physician visits and are related to patient’s interactions with their doctors. Patients often conduct online information searches prior to going to see their physician in order to be better prepared and to enhance their skills for being involved in their health care (Sommerhalder et al., 2009). Patients often express concerns about their physicians’ acceptance of information that patients obtained on the Internet. When they anticipate critical reactions from physicians, patients may conceal their information gathering activities [58]. Patients also engage in information seeking on the Web subsequent to physician visits. Such searches are particularly likely when worries were raised in the appointment and when trust in the doctor is low [22]. In addition to health information seeking relevant to doctor–patient interactions, there are new opportunities for health promotion and patient empowerment created by powerful new health information technologies. Thus, for example, tracking of food or even restaurant and fast food choices has been incorporated in the successful weight management programs currently available through weight watchers. These proactive approaches to health promotion will be discussed in relation to influence of the Web on illness management and health behavior change. An important component of consumer involvement in health maintenance and care involves social support and facilitation including the skill to marshal support (Kahana et al., 2012). The social web can play an important role in this regard. 2.1.2

Personal Health Record (PHR)

The Personal Health Record (PHR) is used to collect, track, and share information about an individual’s health. The intent is to put the person in control of their information, thereby potentially saving time and money. The PHR is increasingly being used by care provider organizations to give patients and providers more insight into a person’s health

288

Health Web Science and the Web of Documents

story. (e.g., indivio). At the same time, cost savings can also accrue to the health care organizations. Aetna — information about doctor visits, fill a prescription, lab test results, then all info about health conditions, R conadded manually or automatically entered and then CareEngine tinually analyzed. It looks at medical, lab, and pharmacy claims and compares to today’s latest medical findings. When an opportunity for better care or when a potential capability of the Web medical error arises an alert is generated. Things such as due for a preventative care visit, or possible health risks. Parents can also keep track of their children’s health records, medications, allergies, conditions, and symptoms, personal information, emergency contact information, health summary, health providers, etc., and can add other medical related information such as over the counter drugs. There are some risks involved in sporadic utilization of health records, particularly by users with limited technological facility. In some health systems those who subscribe to electronic health records, such as MyChart at the Cleveland Clinic, no longer receive mailed notifications about test results.

3 Health Web Science and Social Web

In many ways, the social web is seen as a return to the original values and ideas of the Web as a whole. In contrast to the Web of Documents’ focus on consumption of information and content, the social web brought back the values of sharing, publishing, and expressing ideas from all users in the system. Users are not only consumers but also producers of information and content on the web. The first Web browser written by Tim Berners-Lee was a read/write browser, but was difficult to use. The social web brought back the read/write nature of the original web and thereby supported a new era of web applications that enabled a forum for a collective intelligence1 [35]. O’Reilly’s [163] landmark paper helped to define the emergence of what we call the social web. O’Reilly notes that the social web is focused on having the users of a service actually power a web application. These web applications and services can harness the collective intelligence of

1 Collective

intelligence is a phenomenon in sociology where a shared or group intelligence emerges from the collaboration and competition of many individuals. On the Web, collective intelligence refers to the generation of content that emerges from users networking to generate that content (http://dictionary.reference.com/browse/collective+intelligence).

289

290

Health Web Science and Social Web

users, helping to create a better understanding of what the users want and need. The social web applications build easy to use web services and programming interfaces that are built on hard-to-recreate data sources that become larger and more valuable as more users interact with them. These web services and program interfaces allow for social data sources to be used by other software programs. They enable rapid integration of data sources that produce results that are beyond the reason the raw source data were produced, resulting in a “mashup” culture emerging on the Web (see Section 4.1.5 for a more complete description of “mashups”). The social web applications are also oriented around users and feature technologies that mimic the functionality and feel of traditional desktop applications while being Web-based. We see these technology components in applications that are referred to as the Social web applications such as social networking sites, collaborative filtering, social search engines, blogs, wikis, and mashups. The social web technologies present great promise in the health sciences, healthcare, and health education. Further research and evaluation is needed to best understand how these technologies can be used to help with the wide diversity of health scenarios and objectives. We also note that while Health Web Science evolves and finds kinship with the work conducted by researchers in Medicine 2.0 and HWS, we too acknowledge that there is an ever-growing body of literature in the digital health information and communications technology (ICTs) field, which finds its roots in both communications and media studies disciplines (Stein 2011, Dutta-Bergman, 2006; Street & Piziak, 2001, etc.). The term digital health ICTs conveys a wide range of technologies and applications that increase access to health information and communication, improve immediate and long term health outcomes, and create economic benefits and cost savings for both patients and healthcare providers (Stein, 2011). Stein identifies six unique characteristics and capabilities exclusive to digital health ICTs — immediacy, mobility, multimedia media, ease of search for information, customization/personalization, and time shifting. The use of these technologies ranges from education to e-commerce to self-management tools and include technologies such as blogs, social networking platforms, and interactive television. Moreover, studies on user demographics suggest

3.1 The Social Web Tools in Healthcare

291

that factors such as age, income, ethnicity, race, and health conditions encourage or? discourage health ICT use (ibid). In the next section, we attempt to decipher how different social web applications are currently being used in the health domain and their actual and potential effects. Specifically, we examine the role of blogs, wikis, personal health records (PHR), and health virtual communities. We present a case study of social web technologies used within the context of breast cancer. Lastly, we cover some of the mechanisms found in social web applications and discuss the current research goals and questions in Social Web technologies in health and healthcare.

3.1

The Social Web Tools in Healthcare

Social Networking Tools play increasingly important roles in empowering patients to take active roles in their health care. They may be used for a variety of health related activities. These include information seeking about symptoms as well as diagnosis or treatment options. They allow for creation of communities of patients who share knowledge about of health promotion and maintenance, and exchange ideas about available health care services. Attesting to the broad array of information sharing the “About.com” Patient Empowerment site identifies nine ways that social media tools can help. These include information about the use of blogs and wikis, social networking for personal medical research, information about the use, and benefit of joining patient communities, where to find and how to use health guides, discussion about connecting with and following doctors on Facebook and Twitter. In this section, we review important aspects of social web tools before discussing specific case studies. 3.1.1

Blogs

One of the most common social web technologies is the blog (derived from the term web log). Blogs are easy to use online journals that allow users to publish articles in a reverse chronological (newest posts first) fashion. These articles or posts typically have links to other blogs or web content, and allow the readers of the posts to write and display their comments. Blogs can be linked and communicated together through

292

Health Web Science and Social Web

technologies like TrackBack. This application allows user comment and discussion between and across different blog sites. This communication technology was the basis for the emergence of what many call the “Blogosphere.” Blogs can be found in all areas of health science, healthcare, and education. A search on Google’s Blog Search Engine returned over 156 million blogs with the keyword “health” in them. Some examples of the major health blogs include: (1) the Clinical Cases and Images Blog (http://casesblog.blogspot.com); (2) Life sciences blog (http:// www.lsblog.org); (3) the US CDC’s Public Health Matters Blog (http://blogs. cdc.gov/publichealthmatters); and (4) the US Department of Health and Human Services Health IT Buzz (http://www. healthit.gov/buzz-blog) [55]. In a study consisting of interviews with different bloggers, five common motivations for engaging in this pursuit were found. These motivations include documenting one’s life, providing commentary and opinions, expressing deeply felt emotions, articulating ideas through writing, and forming and maintaining community forums. Although the study did not focus on health, it is easy to imagine that these motivations could also influence patients to blog about their personal health [33]. There has been discussion on the use of blogs not just as an information-publishing tool, but also as a means to help patients become more active in their own healthcare. Adams [2], notes that as patients become more proactive in their healthcare and treatment, they need tools and methods to help record, manage, and organize, information pertaining to their healthcare. Adams discuses that blogs may be a good tool for meeting this need and describes different blogs that do this. For example, “Valtaf” (“losing weight”), a Dutch weight loss site that allows users to keep an online journal about their weight loss program. Users can submit data about their daily diet and exercise progress, weight, fat percentage, BMI, and physical measurements. Users are also encouraged to post personal photos of their progress and write journal entries about their daily struggles. Adams comments that there needs to be more research into the affect blogging has on the health of its users, and how it may affect the views that patients have of

3.1 The Social Web Tools in Healthcare

293

both themselves and others. In the US, for example, Weight Watchers Online provides a program where subscribers can follow step-by-step Plan Guides in conjunction with online tools to track food and exercise, chart progress, find recipes, and workouts. The program began in New York in the 1960s with weekly in-home meetings to discuss weight loss. Today the in-person weekly meetings are complemented with their on-line program, including a blog, to “get the best of both worlds.”2 3.1.2

Wikis

A wiki is a website that enables users to edit the text on the website through their web browser. Wikis were created by Ward Cunningham in 1994 and made live on the Web in 1995 to facilitate the exchange of ideas among computer programmers. Cunning chose to call his first site the WikiWikiWeb from the Hawaiian word for “quick” because it sounded better to him than quick-web.3 Wikis are often used to help in the collaborative process of creating, editing, and publishing knowledge resources. The most prominent features of a wiki are fast and easy editing (through the use of some simplified markup language), versioning capabilities (to help keep track of the history of edits on the content in the wiki and allow for the backup of said content), and article discussions (where editors of the wiki can organize and discuss the issues of the content of the article) [41]. Due to their ease of use and collaborative functions, wikis have become popular in the health domain to help create and publish health knowledge repositories [35]. Some of the more prominent health wikis include Ganfyd (a medical reference wiki that is edited and maintained by both medical and non-medical professionals), WikiProteins (a wiki used to help build community driven knowledge base about proteins), and Flu Wiki (a wiki to help local public health officials and communities to monitor and prepare for influenza pandemics). We make the distinction between wikis and the health virtual community, which we discuss later, because unlike virtual communities wikis lack the same level of interactivity. A major distinction is that 2 http://www.weightwatchers.com/plan/index.aspx. 3 http://c2.com/doc/etymology.html.

294

Health Web Science and Social Web

wikis provide is a simple web-based interface that enables individual content creation on a community (public) site, rather than the creation of a community. With wikis, individuals co-create a shared resource that is an artifact, and some of these artifacts are health information resources. Wikipedia, one of the Web’s largest online encyclopedias, also has a large number of articles on health science and health care. Altmann [7] explored Wikipedia to see how medical informatics was represented. Altmann searched for many different terms used in medical informatics and found that although MediaWiki (the software that runs Wikipedia) was practical and easy to use, key topics in the field of medical informatics were clearly missing from Wikipedia. While Wikis are easy to use and edit, there have been concerns that wikis may contain false or misleading information, and may be easy to vandalize if certain technological precautions are not put into place. This is particularly important in the case of health information and content, which has very high editorial standards. It has been suggested by some researchers that only experts in the field should moderate health wikis. On the contrary, perspective from the Wisdom of the Crows would provide additional support for crowd-sourced information. 3.1.3

Health Virtual Communities

The term “virtual community” is usually used as a blanket term to cover many online communities and social applications. It refers to any online or virtual space in which there can be peer-to-peer sharing of ideas and content. This is usually seen in the form of online forums, discussion boards, and social networking sites. Such fora may encompass a broad array of health conditions, represented by websites, such as “Patients Like Me” where patients can share experiences about the effectiveness of both traditional and alternative treatment modalities in relation to diverse health problems. This is noteworthy, as it has allowed patients to create health communities relevant to even rare conditions. Health related virtual communities are seen in both general-purpose social networking sites (like Facebook) and in communities explicitly designed for a health area, such as My Cancer Place. The latter is a social network to help connect people with cancer to

3.2 Social Web Case Studies in Health

295

help in sharing ideas and experiences. Alzforum is a virtual community that helps with collaboration in the Alzheimer’s Disease (AD) research community. Addressing a constituency of caregivers, CarePages offers a virtual support community for patients and their families. The current literature on the effects virtual communities and social networks sites have on its users is still inconclusive. Some studies have found that participation in online health communities can have a positive effect on people with chronic illnesses [154]. Other studies have found that although virtual communities may help with patient empowerment and organization of treatment, there is insufficient evidence on positive clinical results or effects [68]. Others argue that the value of social network and virtual communities may be outside what is traditional measured as clinical outcome [113].

3.2

Social Web Case Studies in Health

Consideration of the potential usefulness of the Web for impacting health, illness experience, and health care for groups of patients dealing with prevalent serious illnesses offers useful insights into multifaceted contributions of web-based technologies. Such technologies and their socially shaped adoption can help patients cope with challenging adaptive tasks posed by the illness. We now turn to a case study of breast cancer patients/survivors to illustrate some of the uses of the Web for addressing this feared and life threatening epidemic for women throughout the life course. 3.2.1

A Resource for Breast Cancer

Breast cancer offers an excellent arena for a more in depth exploration of patient focused understandings of Internet use [15]. It is a major public health concern, as the most common form of cancer and the second major cause of cancer related deaths in the US among women (American Cancer Society, 2007). In the age of the Internet, a new role has become available to patients as information managers. Information acquisition through the Internet can help develop patient competence in dealing with challenges of life-threatening illness such as breast cancer (Seckin, 2010). Indeed the penetration of the Web in lifes of women

296

Health Web Science and Social Web

with breast cancer is illustrated in research exploring groups related to breast cancer on Facebook. As of 2011 there were 620 breast cancer groups on Facebook comprised of over 1 million members (Bender et al., 2011). This large scale exposure to web-based information offers a great resource. At the same time, there are dangers of obtaining misinformation as well. Not only has the Internet been influential related to health care, but information obtained on line also influences screening decisions. Thus, for example, research by Jorgensen and Gotche (2004) documents the variable quality of information obtained from websites sponsored by different consumer and advocacy groups. Accordingly, some websites omit potential harms of over-screening and most websites appear biased in favor of screening. The increasing sophistication of Internet sites enables patients to access not only sites designed for patients, but also peer reviewed scientific articles that describe the latest research relevant to specific problems of the patients. In addition to gaining specific illness related information, patients also benefit from participation in online communities that offer social support and enhance communication among persons who have similar health concerns. Furthermore, health care organizations, such as those focused on cancer, can relay individually tailored messages to patients who can respond to these at their convenience. Access to information on the Internet has also been shown to enhance health-promoting behaviors. In the case of breast cancer survivors, information about modifiable risk factors has shown promise (Norman et al., 2007). Women diagnosed with breast cancer have played a unique and important role in creatively utilizing the Web to meet challenges of their illness. It has been argued that these women face substantial barriers as they try to make sense of their illness in a fragmented and limited information environment (Hesse et al., 2008). A substantial proportion of women diagnosed with breast cancer are relatively young, well-educated, and proactive patients (Fogel et al., 2002). They represent the demographic group most likely to use the Web for marshaling informational, emotional, and instrumental support. Breast cancer survivors are also a distinct interest group, often identifying as members

3.2 Social Web Case Studies in Health

297

of an advocacy and self-help community. Consideration of the multifaceted uses of the Web by this group provides an excellent case study of the full spectrum of opportunities and also of some of the limitations of eHealth from the patient’s vantage point. National surveys of women with breast cancer reveal that information gathering is the most frequent health relevant function of the Web for this group (Atkinson Saperstein & Pleis, 2009). Additional health related web uses include: communication with health care providers, exploring alternative treatment options, purchasing medications and participation in web-based support seeking, and advocacy. These diverse activities exemplify the four major health relevant applications of the Internet specified by Eysenbach [66]: (1) content, (2) community, (3) communication, and (4) e-commerce. To the extent that breast cancer patients are actively involved in all four of these uses, this group offers an exemplar of research related to the expanding uses of the Web by patients. 3.2.1.1

Information seeking and web content

Women diagnosed with breast cancer have wide ranging information needs. In a study of Internet savvy younger women (under age 45) diagnosed with breast cancer, results suggest that these patients search for information to help them make good treatment related decisions, to learn about their future care and prospects and to pursue social support (Balka, Krueger, Holmes & Stephen, 2010). These goals are congruent with principles of patient empowerment and involvement in health care decision-making. However, it is noteworthy that users of this new resource express significant doubts about the trustworthiness of information they obtain [199]. They also continue to seek information from other traditional sources, such as friends and family. Information seeking on the Web can become particularly valuable, after formal treatments linking patients and health care providers have concluded. The Web may be viewed as both serving information needs and creating needs for further more personalized and tailored information. In many cases, patients experience pressure and uncertainty, rather than true empowerment when faced with a daunting array of information and

298

Health Web Science and Social Web

choices, without sufficient personalized guidance (Balka et al., 2010). Research on the widespread use of information gathering on the Web by breast cancer patients has also challenged accepted notions regarding the digital divide, by demonstrating that socio-demographic characteristics play only a very minor role limiting information seeking by breast cancer patients (Wietfeldt et al., 2010). 3.2.1.2

Breast cancer community and the social web

The use of web-based support groups comprises a relatively small proportion of Internet use by women breast cancer survivors (Atkinson et al., 2009). Thus, it is noteworthy that there is extensive research on this topic. Both medical and social science researchers are fascinated by the role the Internet can play in fostering community among patients coping with the same illness. Peer support can play an important role in helping with the emotional management of this life-threatening illness (Kahana & Kahana, 2010). For example, the analysis of the content of online breast cancer discussion groups indicated that the knowledge gained through peer contact in online support groups provided was an opportunity to access a valuable resource for practical and experiential knowledge [17]. It is important to acknowledge the diversity of web-based programming that is subsumed under the label of community. There are peer-to-peer programs, bulletin boards, moderated as well as unmoderated online support groups. Participation in support groups can offer opportunities for self-disclosure, helping by offering of support to others, emotional release, and empowerment (van Uden-Kraan et al., 2008). Participation in online communities can, at times, also offer misinformation, causing further stress when peers show advancing disease (Salzer et al., 2010). The fascination of researchers and the proliferation of qualitative studies of cancer support groups may attest to the observation that intimacy may be established and “lived out in the disembodied world of cyberspace” (Sandaunet, 2008). It is argued that the disembodied nature of cyberspace may be protective and enabling for patients to enact their identities (Bar-Lev, 2008).

3.2 Social Web Case Studies in Health

299

At the same time, qualitative researchers offer insight into the cultural expectations of sharing socially desirable and positive stories without intimations of underlying chaos in the experiences of cancer patients (Sandaunet, 2008). This line of inquiry is also at variance with beliefs in the therapeutic value of expressing negative emotions by persons facing stressful illness situations (Lieberman & Goldstein, 2008). However, it supports research emphasizing the benefits of expressing positive emotions in the context of the illness experience (Han, Shaw, & Hawkins, et al, 2008). Interviews with Dutch patients participating in online support groups generally confirm empowering aspects of group participation, including improved acceptance of the disease, enhanced self-confidence, and confidence in physicians and treatments (van Uden-Kraan, Drossaert, Taal, Seydel, & van de Laar, 2010). 3.2.1.3

Communication between patients and their doctors

New opportunities in eHealth communication between patients and their doctors have been increasingly recognized as holding both great promise and potential challenges (Kreps & Neuhauser, 2010). Yet, the use of the Web for health communication with providers among cancer patients is relatively infrequent and there is very limited research literature on this subject. The absence of studies would be difficult to attribute only to low levels of utilization of this readily available web technology. It is also at variance with the large body of literature on support groups. A likely explanation rests with the methodological challenges in gaining access to exploring clinical doctor–patient interactions. Studies that do exist demonstrate the potential of the Web for health communication in experimental conditions. Exemplifying these approaches, the value of online health consultation has been found to be acceptable and useful by low-income women with breast cancer (Lu, Shaw & Gustafson, 2011). This online consultation service involved providing computer training and access and resulted in improved doctor–patient relationships. It is important to note that the consultation service did not enable patients to communicate with their

300

Health Web Science and Social Web

own physician and shared some components with uses of the Internet for information retrieval. An innovative approach to doctor–patient information sharing on the Internet is exemplified in the success of the Open Notes project [51]. Another study considered the content of patient communications with doctors on a physician run Q&A board in Korea. Most of the 1355 requests analyzed were seeking informational support (93.5%). Questions were almost evenly divided between information regarding treatments, those focusing on physical condition and those involving lifestyle and self-care. Although the majority of requests (63%) reflected concerns and worry, very few direct requests were made for emotional support. The absence of research exploring web based communication between women who have breast cancer and their personal physicians, attests to the limited use of direct health communication facilitated by the Web as well as ethical concerns of researching privileged doctor– patient communication.

3.3

Patient Empowerment

Health systems have to date have been modeled on delivering care to patients with the professional at the center. “No decision about me, without me” is one of the headlines arising from the White Paper Equity and Excellence: Liberating the National Health System (NHS) that envisages an NHS that is outcome focused, clinically led, patient centered, and in keeping with NHS core values [54]. This is in alignment with the Salzburg Statement on shared decision-making [61] that recognizes the role of the patient as an equal participant in their care. The King’s Fund, a charitable foundation in the UK that seeks to improve health and health care in England, subsequently produced a document to clarify the term “shared decision making” and suggest the skills and resources required to implement it, as well as outlining the action needed to make the vision a reality [45]. All three documents suggest a top-down hierarchical approach to healthcare and do not consider “the wisdom of the crowd approach” which has begun to find traction in industry and business to shape service delivery [75]. With

3.3 Patient Empowerment

301

the Internet as a conduit for health care delivery, new possibilities for patient empowerment have emerged. Roberts [174] concludes “whilst there is no consensus amongst analysts regarding how best to define ‘patient empowerment,’ she states that at the very least, this concept entails a re-distribution of power between patients and physicians where empowered patients attempt to take charge of their own health and their interactions with health care professionals. Furthermore, patients have different ideas about what it means to ‘take charge’ and ‘be empowered’. Ten years later, van Uden-Kraan et al. [206] summarized the literature. Empowered patients were then described as ‘patients successful in managing their condition, collaborating with their health providers, maintaining their health functioning and accessing appropriate and high quality care’. They conclude ‘empowerment reflects the belief in patient autonomy and the right and responsibility of patients to access health information and to make their own health related decisions’. Implicit in this is patient choice. Saltman [182] warns that offering choice can either strengthen the citizens commitment to traditional welfare objectives or alternatively, it can severely damage that commitment, depending upon the design of the choice mechanism and the structural context within which patient choice occurs. For patient choice to be linked to true empowerment, choice must reinforce rather than undercut the accountability of health care providers to the population they serve. Regardless, empowerment is a multifaceted concept that has both an enabling process and an outcome [9]. Some empirical evidence suggests that active patient participation in health care is associated with better patient outcomes. This field is fertile for future studies which both help to develop theoretical models of patient empowerment and articulate the conditions under which patient empowerment occurs [174]. The Patient Empowerment Network (PEN, http://www.powerful patients.org/wp/) succinctly summarizes the use of the Internet to aid patient empowerment. “There are many, many websites that provide information about specific diseases, about medical research, about physicians and medical centers of excellence, and about treatment choices. Our goal is to support you, your family and your friends to

302

Health Web Science and Social Web

navigate through all this information and use it to make the very best decision you can about your own particular health concern. . . to put you in the “driver’s seat.” We’ll do that by providing online content, live events, and information delivered in ways we haven’t thought of yet, as well as opportunities to exchange information with others who are dealing with the same medical concerns you are — and by creating a network of support. Cumming et al. [47] show how the use of digital story telling potentially empowered women who accessed a menopause website to actively seek help for vaginal atrophy — a subject that they had found too embarrassing to speak to a health professional about before viewing the digital story. 3.3.1

e-Patient Dave’s Story

David deBronkart, described as an “e-Patient,” by Health Leaders magazine’s cover story in Sept 2009 “Patient of the future,” is an advocate for patient rights and self-advocacy in the patient role. As a supporter of patient rights, he is an ethicist and a technologist. He advocates for the strengthening of participatory medicine and personal health data rights. e-Patient Dave’s history is correlative to the policy and thought of Health Web Science. In 2007, as a result of a routine shoulder X-ray, Dave deBronkart was diagnosed with stage-four renal cancer. With tumors in both the lungs, and metastases to bones and in muscle tissue, he was given a median survival of 24 weeks. He received a treatment at Boston’s Beth Israel Deaconess Medical Center that usually doesn’t work, but for him it did in six and a half months he stopped treatment and seems unlikely that the disease would reappear [49]. The whole story is in his first book, Laugh, Sing, and Eat Like a Pig, and is told in videos of his speeches at ePatientDave.com/videos. Dave contends that if he were not so engaged he would not be alive today: After my illness I found myself “tumbling down an Alice’s Wonderland rabbit hole, becoming “e Patient Dave,” blogging, suddenly testifying in Washington, and now giving speeches about how empowered,

3.3 Patient Empowerment

303

engaged, activated “e patients” are changing what’s possible in medicine. But I’ve learned there’s a problem — a big problem — and that’s what this book is about. The problem is that our culture assumes doctors know everything and patients can’t possibly add anything useful. It’s not just doctors who think that — most patients do, too. Because of the cultural disconnect, most patients don’t speak up. And when activated patients do, it’s often sneered at, or politely dismissed with a “stay off the internet.” (This isn’t universal; some doctors and nurses “get it.” But if you yourself haven’t experienced being put in your place (at least gently), ask your friends and neighbors. I’ve never seen a case where stories of disrespect are more than one person away from any individual.) However traumatizing, these events were not the impetus for e-Patient Dave to advocate for participatory medicine and patient data access. Dr. Danny Sands asked Dave deBronkart to join the annual retreat of e-Patient Scholars Working Group, Founded by the late Tom Ferguson MD. Nearly a year after his diagnosis, this became the impetus for e-Patient Dave to start to advocate for participatory medicine. Dave deBronkart became e-Patient Dave advocating for patient rights, especially when it came to technology. A widely cited aspect of e-Patient Dave’s case is his frequent use of the Internet during his illness. He joined an expert patient community at the Association of Cancer Online Resources (ACOR) and he sought information about his disease on various websites. Dave also shared access to his medical record in the hospital computer systems, as he sought advice from medically knowledgeable family and friends. He emphasizes the key role played by the access to his full medical data in order to obtain life saving advice and treatment. Independent of the Web.

304

Health Web Science and Social Web

Here’s reality: • There were 800,000 new articles published in medical journals in 2010 alone, and the number keeps going up. • Paul Grundy MD, IBM’s director of worldwide healthcare transformation, told me the average primary care doctor has 1500–2000 patients. Imagine how many conditions s/he has to keep up on.4 • My own oncologist, Dr. David McDermott, says it’s true: “Unless you’re a sub-sub-specialist like me, it’s impossible to keep up.” Clinicians have information overload, but the Internet lets ordinary patients see information that their doctors might not. Yikes: does that clash with our cultural assumption! Moving beyond his own health care he became a recognized advocate for other patients as he started an online journal and support community on CaringBridge. E-Patient Dave is a champion of participatory medicine and personal data access rights. Participatory medicine is a phenomenon in which patients are fully engaged with health care providers and in being self-advocates and decision makers in their health. A key facilitator of effective advocacy is based on personal health data rights. When the patient is in control of his or her health, data tools exist for effective self-advocacy. 3.3.1.1

An analysis of e-patient dave’s case

A research question that the e-Patient Case poses is whether or not e-Patient advocacy is efficient, to elaborate, whether as total work expended by all parties is increased whether or not there is a greater chance of success in treatment or diagnosis. Certainly all patients and doctors have rights, but the ethics of the situation call for a deep understanding of what each action means to the inalienable rights of

4 deBronkart’s

reference [9] “I [deBronkart] recently heard there’s a list of 11,000 conditions, but I haven’t found the source.”

3.3 Patient Empowerment

305

humankind, for instance like those in the second bill of rights. Certainly the case has brought to the forefront questions about policy that were previously nonexistent. 3.3.2

Harmful Communities in the Social Web

In the previous sections, we have discussed many of the positive effects the web has had in health sciences and healthcare, but this is not always the case. The Web, with its ability to connect people all around the world over the most esoteric subjects can also be used for proposes that can be harmful. In this section, we discuss some case studies of communities that promote unhealthy life choices and the effect it had on their members. We also discuss the importance of privacy in terms of health data on the Web. The Web allows for instant access of a wide diversity of data and information, but in some cases people may wish to keep personal health information confidential. We discuss the current state of health privacy on the Web, and the potential dangers of the loss of privacy. Ultimately, Health Web Science seeks to create knowledge, tools, and know-how that enable, empower, and motivate the formation of good communities and enable discouragement of bad communities. Lessig (2006) — Cyberspace would thus change to take on Code and commerce, the two powerful forces of social order characteristics favorable required to build a well-regulated world. 3.3.2.1

Pro ana communities

The term Pro-ana has been used in recent years to offer a positive outlook on and to promote the eating disorder anorexia nervosa. On the Web many virtual communities have popped up to help support this community. Some of these communities use popular social networking sites like Twitter, Facebook, and MySpace, while others have been built just for the pro-ana community. Researchers have looked at and analyzed the community of sites that make up the pro-ana community and the effect it has had on its members. What we found was data about the specific features or aspects, and some about motivation, but little about the role the Web plays.

306

Health Web Science and Social Web

Many of these pro-ana communities share common features and attributes. The most common feature among these sites is the endorsement of anorexia as a positive lifestyle choice. A survey by Borzekowski et al. [34] showed that 84% of these sites have this in common. Many sites offer strategies for ways to hide eating disorders under more socially acceptable guises like dieting or veganism [160]. Many members of these communities compete with each other on their weight-loss and congratulate each other after a fast or binge [176]. Members also share personal information like their measurements, weight, pictures, and experiences of going through different diets [188]. According to the same survey, 83% of these sites share crash dieting recipes and skills. Members of these sites post information about how to best hide the side-effects of anorexia from doctors and family members, ways to ignore hunger pangs, and techniques on inducing vomiting [102, 188]. About 84% of Pro-ana sites also heavily feature what is referred to as “thinspiration” [34], which consist of media that encourages and helps to motivate losing weight to look thin. Due to the controversial nature of the content of these sites, and the threat of being shut down by online providers (George, 2002), many pro-ana sites (58%) have taken measures to hide themselves or claim to be pro-recovery [160]. Some sites even interview and screen potential members to limit exposure [80]. There has been some research into the motivations behind these pro-ana communities and their users. Some studies cite that these communities function as a coping mechanism for those with anorexia (Mulveen et al., 2006). In a survey by Ransom et al. [171], pro-ana members reported that a sense of belonging, social support, and support for helping with current eating disorders were reason for joining a pro-ana community. Support with stress, meeting other anorexics, and finding triggers for anorexia were reported as motivation to staying with a pro-ana community. Research into the effects of participating in a pro-ana site has shown that 84% of members on average decrease caloric intake by 2470 calories per week [116]. A total of 54% of members though are able to perceive this reduction. From the same study, 24% of participants continued to use the weight-loss techniques acquired from pro-ana sites three weeks after the experiment. Studies that controlled for viewing pro-ana sites

3.3 Patient Empowerment

307

and viewing non pro-ana sites reported no decrease in caloric intake. Other studies have shown that pro-ana sites can negatively impact selfesteem and self-image [15, 16]. Some studies have looked into ways to handle pro-ana sites. While shutting down and censoring pro-ana sites have seemed to only push the communities more underground, some research has looked into alternatives through education and awareness. Using a click-through warning on the dangers of anorexia before entering a pro-ana blog, 33% of the visitors ended up not continuing on, but the number of pro-ana blogs actually increase 10-fold with two times the amount of traffic to such blogs than was recorded at the beginning of the study (Waarschuwing 2008). The continuing study of this movement has revealed a lot about how disorders can be encouraged and exasperated on the Web, but little information so far about what can be done to help healthcare providers develop better strategies in how to treat this disorder, or in empowering patients to recognize that seemingly desirable lifestyle choice may be harming them. 3.3.2.2

Bug Chasing and Gift giving Movement

Bugchaser is the slang term for individuals who seek HIV positive sexual partners for the purpose of becoming HIV positive through unprotected sex. Gauthier and Forsyth (1999) reported that with the increasing presence of the World Wide Web, “a forum now exists for providing links between like-minded individuals in pursuit of this goal.” Tewksbury et al. analyzed the attitudes, behaviors and demographics of on-line bug chasers as has Gov et al. (2006), who analyzed 1228 online profiles and found six categories. These categories included committed and opportunistic bug chasers, committed and opportunistic gift givers, Serosorter (those who did not match with the bug chaser/gift giver identity), and ambiguous bug chaser or gift giver. Studies into the motivations for gift giving or bug chasing have found to be very diverse. Blechner et al. (2007) found that bug chasers were typically alienated and viewed HIV as a path to being part of a sympathetic community. Others engaged in bug chasing as a relief from

308

Health Web Science and Social Web

the anxiety of contracting HIV. Moskowitz et al. [151, 152] viewed bug chasing as a form of sexual addiction, with the high-risk behavior serving as a type of high. A study from LeBlanc et al. [127] agrees more with this interpretation. The study, which consisted of a survey to members that self identified as bug chasers, reported the highest motivation found was the view of becoming infected with HIV as erotically thrilling. These case studies of potentially harmful Internet-based patient outreach underscore that the Web offers tools that are generally used to benefit human health and well-being. However, ultimately it is up to the users to ensure that ethical uses prevail and that tools are not misused to create harmful health outcomes.

4 Health Web Science, the Semantic Web and Linked Data

In the previous sections, we have noted how people interact and use the Web. We see that, although the Web works well for humans, computers have a very limited understanding of the information on the Web. All of the knowledge and information on the Web is expressed in human languages. Human languages are ambiguous and thus difficult for computers to understand. Computers understand the syntax of the Web; they see that one document or web page with a certain web address, known as a Uniform Resource Identifier (URI) links to another document with a different web address, but machines do not understand the meaning behind these documents, or why these documents are linked together. What they lack are the ‘semantics’ of the content of the Web, like language, has semantics so that speakers of the language can understand each other. The Semantic Web can be understood as a new World Wide Web, one which focuses on meaningful use and both human and computer understanding, in short, a web of concepts. This next generation of the Web built on top of the existing World Wide Web technologies, which helps describe to both humans and machines the meaning of web content. In short, the Semantic Web is a way to help computers understand and use the Web in a manner 309

310

Health Web Science, the Semantic Web and Linked Data

Fig. 4.1 Information Flow in the PopSciGrid community health portal.

akin to how humans do (see Figure 4.1). On the other hand, Linked Data, is what its name implies. The data that the World Wide Web is stored on may include a number of formats, such as SQL databases, flat files and XML, the eXtensible Markup Language. These data are generally “siloed” or separate and non-interactive to each other, meaning that the data of an application that is made in a siloed manner is only applicable to that application. However, if the data is linked then the data of the same application is not siloed only to that application, this is Linked Data; because of this, Linked Data has the ability to build upon itself [146]. For instance, one application is particularly good at describing a certain kind of event, and another is very good at describing a setting for this event, in a Linked Data modal these two data sources would be able to mesh at some level to wean which events where at which settings. These interactions compound on each other and in some modals lower level logic can be introduced. Linked data is not the Semantic Web, they are not mutually exclusive and there is some amount of argument about whether Linked Data encompasses the Semantic Web or vice versa. In practice, this means that the information in the Semantic Web is far more granular and far more explicit — machines cannot rely

4.1 Semantic Web Basics

311

on intuition to understand content on the Semantic Web. Data and concepts on the Semantic Web are all explicitly named, and the links between them have specific types (like “involved in” from Figure 4.1). As such, machines can now begin to “understand” the information within a Web page, and precisely how that information relates to information in other Web pages.

4.1

Semantic Web Basics

The Semantic Web is a combination of two somewhat inter-dependent concepts — Linked Data, and Semantics. At its core, Linked Data requires: that individual elements of data on the Web should have a unique address (most often a URL); and that these addresses, representing data elements, should be connected to one another using explicitly labeled links that describe the relationship(s) between those data elements. Semantics means that there should be a commonly understood, shared, and explicit definition for every kind of “thing” that might be found on the Web (e.g., genes and diseases, as in Figure 4.1), as well as the kinds of relationships that might be found between these things, and exactly what those relationships “mean.” Semantics are encoded and published in a variety of ways, but generally speaking these shared definitions are described in a knowledgestructure called an Ontology, the word ontology here is used differently from its traditional meaning, here its use as a cognitive and computational tool is emphasized over its meaning as a philosophical concept of “The Ontology.” For example, one of the most popular controlled vocabularies in Linked Data is “FoaF” or Friend of a Friend. This vocabulary helps describe people and the different attributes, or properties; one might want to use to describe people as well as the relationships between people. Putting these ideas together, using Figure 4.1 as an example, we see that the data element BRCA1 has a web address, and that the data element Breast Cancer has a web address, and that those two data elements are linked by the “involved-in” relationship. This is an example of Linked Data. Moreover, we see that the data element BRCA1 has a type — gene — and that the data element Breast Cancer also has

312

Health Web Science, the Semantic Web and Linked Data

a type — disease. The concepts of ‘gene’ and ‘disease,’ along with the exact meaning of the “involved in” relationship, will all be defined by an ontology published somewhere on the Web. Thus, when a machine encounters BRCA1 on the Web, it is able to follow the “type” link, discover that it is a gene, then automatically look-up what “gene” means, and thereby make decisions about how to treat that piece of data (for example, what kinds of analytical tools operate on “genes,” or what kinds of properties genes have, such as DNA sequences). Thus, Linked Data and Semantics work hand-in-hand to create an overall global network of machine-readable information known as the Semantic Web. Finally, if the Linked Data are made publicly available, and follows a few simple best practices (http://linkeddata.org), the resulting information becomes part of the Linked Open Data (LOD) graph (http://richard.cyganiak.de/2007/10/lod/, http://inkdroid.org/ lod-graph/). One well used description of LOD comes from Wikipedia as “a term used to describe a recommended best practice for exposing, sharing, and connecting pieces of data, information, and knowledge on the Semantic Web using URIs and RDF.” Tim Berners-Lee [28] in his linked data designs document defined four basic principles for linked data. (1) linked data should use URIs as names for entities, (2) linked data should use these URIs so that people can look up the name (and dereference the names), (3) when a URI is dereferenced, useful information should be provided using Semantic Web standards, and (4) linked data should have links to other URIs to enable better discovery. The colored version of the Linked Open Data cloud diagram (http://richard.cyganiak.de/2007/10/lod/lod-datasets 201109-19 colored.html) groups the data sets by domain and provides a visualization of the many Health and Life Science data resources that provide their data according to LOD guidelines. The benefit of the Semantic Web is that it attempts to model the information and knowledge on the Web for computers to use, allowing computers — unaided by humans — to query and reason over this information. In the original Scientific American article introducing the idea of the Semantic Web [29] a lot of focus was put into the idea of software agents that would make decisions and interact on the Web in much the same way that people do. As Semantic Web technologies

4.1 Semantic Web Basics

313

grew and matured over time, the ideas that emerged focused less on software agents, and more on linked data and vocabulary reuse. One of the biggest immediate advantages of the Semantic Web was seen in its ability to link heterogeneous data sets and structures together in a more bottom up, decentralized and web-like way. Though fully automated exploration and analysis is still emerging, this grand vision no longer seems implausible as adoption of the Semantic Web becomes more prevalent, particularly within the health and life sciences. 4.1.1

Semantic Web Technologies and Languages

The Semantic Web idea is complex. Its facets and ideology are not easily described or defined beyond the core concepts of frameworks and meaningful data. Furthermore, it is a moving target to define with ideology or methods. To understand the Semantic Web better, it is useful to have an understanding of its popular ideas and technologies. The fundamental technology of the Semantic Web is the Resource Description Framework, or RDF. The RDF defines a syntax for modeling knowledge and data as entities and relationships, meaning that syntax has a technical tool that defines it. In RDF, Uniform Resource Identifiers (URIs) identify both the entities and the relationships. The concept of a URL as a unique identifier is not out of the question, domain names are unique to websites, and domain names are URLs. A Uniform Resource Identifier (URI) identifies a resource either by location (URL), or a name (URN), or both (UEL). More often than not, most of us use URIs that defines a location to a resource, the URL is commonly known as a web address. If, say, a URL is being used as a URI, then it is being used to represent some concept, say a name, or a relationship. Assigning that it is a URI, means that we have a point, in a graph. Using another URI as a relationship means that we can connect the points. Very soon a web starts to appear. The most basic data-structure that can be represented in RDF is the Triple, consisting of three parts: the subject (the concept being described), the predicate (the named relationship connecting the subject to another concept), and the object (the concept that the subject is linked to by that relationship). Think of the triple as a simple

314

Health Web Science, the Semantic Web and Linked Data

sentence that states a fact or idea. Using Figure 4.1 as our example again, the relationship between gene and disease is in the form of the Triple [BRCA1] — [involved in] — [Breast Cancer]. The only difference in RDF would be that instead of using actual English words, which are ambiguous and non-unique, the URIs of these data-points are used to construct the “sentence.” The “Web” aspect of the Semantic Web not only refers to the fact that these triples are published on the World Wide Web, but also that they are inter-connected, where the object of one triple becomes the subject of another. For example, in Figure 4.1, breast cancer is the object of the triple relating it to the BRCA1 gene, and it is also the subject of the triple that describes the tissues affected by breast cancer. These sometimes highly inter-connected networks of triples are known as “Graphs,” where the meaning of “Graph” here differs from its more common usage in health science. By expressing data and information as an RDF graph it becomes more straightforward to link or connect different data sets together simply by creating predicates that join a subject from one data set to an Object in another data set. A particularly powerful example of this idea is demonstrated by the RDF solution to the problem of “synonymy” — where two data sets have different names for the same “thing.” On the Semantic Web, this is commonly solved by creating a “same-as” triple between the equivalent concepts in the two data sets. This allows machines to then bridge the data sets, creating a more complete graph of knowledge by combining the information from each source. For those familiar with relational databases, this is very similar to doing a “join” between two tables, but importantly, this “join” is happening between data resources anywhere on the Web, and is entirely independent of, and unaffected by, the underlying database software or schema. As such, the Semantic Web is an extremely flexible way to merge data from disparate and non-cooperating/coordinating resources. Ontologies and Vocabularies are constructed using a variety of languages, with various levels of complexity and expressivity. Resource Description Framework Schema (RDFS) provides some basic constructs for defining the vocabularies intended to be used by machines to interpret RDF statements. The RDFS vocabulary includes the idea that

4.1 Semantic Web Basics

315

there are “things” (classes); that these classes can have “types,” and properties; that classes can have sub-classes; and that certain properties can only be applied to certain things (domain), or that these properties are restricted to having particular values (range). These constructs also allow us to begin to reason over our data and find new information that was not explicitly there, and/or validate if the data set has been constructed “sensibly.” For example, we can describe the concept of patient and person in RDFS, and also state that all data items classified as a patient can also be classified as a person using the sub-class relationship. By knowing this, if a machine finds an item in a data set that has been classified as a patient, it also knows that the data point is also referring to a person, and thus will have the properties of “personhood,” even though this may not have been explicitly expressed in the data. Although this is a very powerful tool to be able to express these ideas and concepts, there are still a lot of limitations in what can be expressed using RDFS. The Web Ontology Language (OWL) was developed to address this issue, allowing for the creation of more expressive knowledge structures by stating, for example, that genes and proteins cannot be the same thing, and that they must have different properties. As such, invalid “joins” between data sets — where one data set is referring to the BCRA1 gene, and the other referring to the BCRA1 protein, can be automatically detected and rejected, and similarly, that properties of genes cannot be attached to data-points that are known to describe proteins. 4.1.2

Bio/Medical Ontologies

Medical ontologies have existed for quite some time; many have existed even before the Web. In fact, “nosologies” — classifications of diseases — have been used in healthcare for centuries! These can be expressed more precisely using Semantic Web languages like RDFS and OWL, which would allow machines to more fully understand the ideas and concepts in health data sets and reason over these data sets to infer new information that was not explicitly declared. Semantic technologies can also be used to detect logical inconsistencies in the data, for example, as a part of quality control. The spectrum of semantics used

316

Health Web Science, the Semantic Web and Linked Data

in the health sciences is broad, ranging from straightforward controlled vocabularies and terminologies; structured or hierarchical vocabularies where terms become more specific along a hierarchical path; and formal representations of knowledge, including detailed definitions of concepts and the relationships between them. They are also applied to a wide range of functions, including harmonization of diagnostic codes for billing and/or clinical purposes, facilitation of search, data exchange and integration, and formal reasoning about biomedical information for the purposes of automated decision support or heath research [32, 179]. Bodenreider makes an interesting note on the risks and rewards of the diversity of biomedical ontologies. While the great diversity of biomedical ontologies allows users to find ontologies that can best fit their goals, the diversity also means that there will be mapping issues between data sets from groups who have selected different ontologies. Moreover, the differences in granularity of the terms in different ontologies means that mappings will not always be possible since some concepts will not be defined in the less expressive ontologies. This can be a significant challenge unless a bridge or link between the two ontologies has previously been established. Bodenreider notes that multiple large ontologies (like the Unified Medical Language System or UMLS) have been built and are used to help bridge and connect some of the different biomedical ontologies together.

4.2

Why Semantic Web for Health

In the health field, there are a multitude of data sets, published in different formats by different organizations. Furthermore, the data in healthcare and life science are usually not complete because most data sets do not come with enough information (either formally encoded or in natural language text) that describes the context that is needed to properly understand the data. (Cheung et al., 2008) suggest that no single data set can capture the complete context of a given research study. One motivation for linking data sets, therefore, is to help understand the data by providing additional context and meaning that computers can use. This provides the potential to allow additional automated support for validation and consistency checking using the additional

4.2 Why Semantic Web for Health

317

meaning provided by the links. It can give rise to new ideas and theories and the automated checking support can help validate or invalidate old ones. Goble and Stevens [86] also discuss the need for data integration in Healthcare and Life Science and the need for common identifiers, shared semantic and shared access mechanisms and how these will facilitate data integration. The Semantic Web, and the Web of Linked Data in particular can help with this, allowing queries and applications to be built on top of a linked web of health data. There is too much health data for us to deal with on our own. From monitors in hospitals to published research related to social networks, health data is increasing at an exponential rate. This increase in data is often too taxing for human assimilation. We need to increase our usage of machines to help with this information overload. The Semantic Web can take some of the load off patients and health care providers to more fully use the knowledge and data we already have. To help in the effort, the World Wide Web Consortium (the standards body that helps to develop and publish the technology recommendations for the Web) launched the Semantic Web for Health Care and Life Sciences Interest Group (HCLSIG; http://www.w3.org/ 2001/sw/hcls/) in 2007. This interest group has been chartered to help with research and collaboration in the Healthcare and Life Science domain by supporting, developing, and promoting Semantic Web technologies [181]. 4.2.1

Linked Open Health Data

There are an increasing number of players in the space of Linked Open Data for health although they are not necessarily competitors. We mention just a few below to highlight some of the projects that display best practices in creating and publishing Linked Health Data. One of the earliest and biggest players in Linked Open Health Data and a nice pedagogical example was Bio2RDF. Bio2RDF [23] provides RDF serializations of different public bioinformatics databases such as the Kyoto Encyclopedia of Genes and Genomes (KEGG) [120], PDB [25], MGI (Bake, JA, et al., 2003), HGNC [36] and several of the (National Center for Biotechnology Information) NCBI’s databases.

318

Health Web Science, the Semantic Web and Linked Data

The system used different RDF-conversion programs (commonly called “RDFizers”) to transform these different data sets into RDF and published all of these RDF documents using an open source triple store. Bio2RDF thus provides a warehouse of health-relevant data using a Semantic Web approach for integration. Along with converting these different biomedical data sets, Bio2RDF created a URI syntax that is consistently applied to the elements found in the data sets. By normalizing the URIs, Bio2RDF allows for better cross-data set queries. Other significant efforts at publishing biomedical data in Semantic Web formats include: OGOLOD [164], a knowledge base about orthologous genes and proteins; DrugBank [216] a repository of almost 5000 FDAapproved small molecule and biotech drugs; Linking Open Drug Data (LODD) a mash-up of 12 databases relevant to pharmaceutical research [115, Samwald et al., 2007]; and UniProt [14]. While many governments around the world have started to publish their health related data on the Web, these are often published in formats that are difficult to query and integrate. Because of this, a number of government and non-government organizations have begun to convert and publish data sets as linked data. The UK government has been supporting their open data operations with linked data from Day 1. Many of their health census and survey data sets are in RDF and can be found in open access triple stores. The US government’s health.data.gov group has adopted a linked open data approach. For example, in one effort, they have been building a vocabulary to describe hospitals in the US, and using this vocabulary to publish hospital metrics and measurements as linked data. This vocabulary also provides a more authoritative space where other can link health and hospital data to. The health.data.gov group as published the metadata for all their health data sets as RDF in their catalog, allowing for easier ways to mine, use, link and search this metadata. Non-government organizations and academic institutions have been assisting with these efforts. For example, researching the resources required to find, organize, republish, and scale linked data, or by re-publishing governmental data in Linked Data formats. For example, Oktie Hassanzadeh, from the University of Toronto, has published a linked data version (http://linkedct.org) of the US Clinical trial database (http://clinicaltrials.gov). Researchers at RPI

4.2 Why Semantic Web for Health

319

have also converted and published hundreds of US government data sets as linked data (http://logd.tw.rpi.edu), many of which are in the health domain. Faceted search services allow people to find data sets that have been converted to RDF in any topic, including a significant number of health topics. The International Open Government Dataset Search Engine (http://logd.tw.rpi.edu/node/9903), which searches over 1 Million Datasets around the world, finds 5156 different data sets under the term “Health.”1 Finally, Personal Health Records are also starting to adopt Linked Data guidelines and technologies. For example, Caregraf is a system that enables the conversion of data from EHRs and PHRs into linked data, and uses many standard healthcare vocabularies and ontologies. 4.2.2

Mashups and Web Applications

As health and biomedical data continues to grow, there has been a need to build applications to help integrate and visualize this data for end users. These Web applications are typically called “mashups.” Mashups use and combine data, presentation or functionality from different sources to create new services and functionality. Most mashups pull information from open data and/or open analytical interfaces to create novel data sets on the fly. Recently, mashups have started using and being built on top of Linked Data, taking advantage of the already linked data sets in the Linked Open Data cloud. As more health and biomedical data is published as linked data, health mashups are becoming more common. Some examples of semantic health mashups include a mashup that integrates genomic and phenomic knowledge to help find disease-causal genes (Stevens et al, 2006); Semantic Web Applications in Neuromedicine 1 The

International Open Government Dataset Search Engine (http://logd.tw.rpi. edu/node/9903) provides metadata extracted from the catalog websites that can be utilized for faceted search. Using the search term “Health” retrieves 5156 data sets as of April 2013. An alternate filtering approach using health-related topics such as “Health and Safety,” “Health and Social Care,” “Health and Nutrition,” and “Health” retrieves 4508 data sets, of which, 1630 are not included in the first search. As with the early days of the web, metadata is inconsistent across catalogs and within the catalogs themselves. This example illustrates the potential for improving consistency and ease-of-use through standardizing with a common metadata model.

320

Health Web Science, the Semantic Web and Linked Data

(SWAN), a semantically based website for Alzheimer Disease research (http://hypothesis. alzforum.org/swan/do!getHome.action; http://dx. doi.org/10.1016/j.websem.2006.05.006); yOWL, a knowledge base for yeast research [208]; and TrialX, a web application that helps connect patients to clinical trials using PHRs and Semantic Web technologies (Patel et al., 2009). Clearly, the desire to mashup data from a wide variety of sources is greatly facilitated by these resources publishing their data using Semantic Web standards, since these are specifically aimed at facilitating cross-resource linking and query. In developing and using mashups, Cheung et al. (2008) provides a nice summary of their strengths and weakness. As strengths, they list applicability (the many ways mashup tools and semantic data can be used in a wide variety of health domains), ease of use (little to no programming is needed to create and publish some of these mashups), reusability (many mashup tools and data sets are built to be reused for other purposes), and interoperability (many of the tools can be used together even though they were designed and published by different organizations). The weaknesses they list were instability, scalability, security, and uneven quality of final output. While mashups have great promise in the health field by allowing non-technical experts to mix and visualize data, they still suffer from performance issues with large amounts of data, and providing the security needed due to the nature of health data and information. 4.2.2.1

Population sciences grid

In this section, we describe an example of next generation community information portal. We describe a particular exemplar portal and also discuss how it may be indicative of a growing evolution in using semantic technologies and linked data to facilitate access to health data. Population Sciences Grid (or PopSciGrid) project aimed to create a place where users can explore potential relationships between selected health-related behaviors, policies, and demographic data. It also aimed to explore how semantic web technologies and linked data could serve as enabling technologies for next generation health informatics portals. The project was undertaken by a multidisciplinary team of researchers,

4.2 Why Semantic Web for Health

321

with expertise in cancer prevention and control, tobacco, semantic web, and social networks. The portal project mashed up data about smoking prevalence, demographics, and intervention policies such as smoking bans and pricing strategies and then provided interfaces that allowed people to explore smoking prevalence over time and by geographic regions and view this alongside data that showed for example, types of smoking bans (e.g., workplace, restaurant, bar) and when they went into effect, pricing and taxing policies and amounts. Interfaces made it easy to view smoking prevalence and their changes in time periods associated with policy changes in intervention strategies to allow people to see at a glance if smoking prevalence was changing at the same time or just after intervention strategies were enacted. One of the many things that could be seen was that smoking bans co-occurred with lower smoking prevalence. The information flow for the portal is shown in Figure 4.1. The data sources were identified by domain experts, in this case cancer control and prevention experts, tobacco researchers, and relevant policy experts. The data were then converted to linked open data and published in flexible and accessible data stores, in this case an SPARQL end point. Multiple visualization interfaces were produced that allowed easy access to anticipated data. For example, geographic maps of smoking prevalence over time were provided and interactive interfaces were created that allowed users to “play” the data over time — allowing people to see when bans went into effect and simultaneously visualize prevalence and change in smoking data. More recently [141], interfaces have been generated that aim more at statisticians allowing them to peruse data and look for correlations. One of the things that makes this project interesting is that it uses relatively simple semantic technologies to (1) convert the data into linked data, (2) connect the data to other data sets (using relatively simple background ontologies and relatively simple enhancement rules supporting mapping, (3) analyze the data against semantic constraints, and (4) ontology-enhanced queries to populate interfaces. This general workflow has been used multiple times in a wide array of data domains. The same flow has more recently been used to produce portals that allow access to the health data in the HHS health data challenge.

322 4.2.3

Health Web Science, the Semantic Web and Linked Data

Data Availability and Data Science

A differentiation between Health Web Science and disciplines under the Health 2.0/Medicine 2.0 umbrella is one of scale; in Data Science, there is an emphasis on “Big Data.” The term “Big Data” refers to data sets that are too large for traditional data management and analytical approaches. A lot of data exists in information silos (information systems that are unable to communicate with other management systems) thereby potentially limiting productivity and novel solutions. These “Big Data” sets are rapidly increasing due to technological advancements, even from the smaller laboratories. Big Data spans at least four domains: data volume; data velocity, i.e., analysis available immediately, for example, the “health care singularity” [83]; data variety, i.e., data in all its different forms; and data veracity or data provenance, that is the trustworthiness of the information or source. Much work remains to be done to make information in Big Data, accessible including capture, curation, storage, search, sharing, linkage analysis and visualization of the data to allow interrogation to be of maximum value. The quantity and complexity of Big Data necessitate the development of new tools capable of automatically converting this data into new biomedical knowledge that is accessible to the clinical practitioners, researchers, and the public. It is in this context that the emergence of the Semantic Web, and its related technologies, are important, whereby, machines can explore vast networks of inter-connected data in a meaningful and computationally efficient way, bringing in knowledge distributed around the Web in order to elucidate the new emergent knowledge inherent within such data networks. The quantified self (http://quantifiedself.com/about/) (know thyself through numbers), whereby a person collects his/her own personal data from sensors/smartphones and uses this information as a feedback loop for self improvement, is gathering momentum and resonates with the wellness agenda and the need to empower people to look after and take responsibility for their health WHO [214]. Thus, the quantified self is a natural progression from the current practice of monitoring

4.2 Why Semantic Web for Health

323

patients either acutely in real time or as part of managing a chronic illness. Using the principles behind Big Data emergent properties and novel solutions may arise. Smart/intelligent Cities (http://www.smartcities.info/) (using information communication technologies to create networked infrastructure across the economy, mobility, environment, people, living and governance) and Smart Homes [149] (where houses are technologically enabled to assist independent living) will increase the amount of data available for analysis and interpretation and novel solutions. Due to the size and complexity of “Big Data,” particularly in the era of P4 (personalized, preventative, predictive, participatory) medicine (e.g., personal genomes, the quantified self, data from environmental sensors in Smart Cities) clinical decision-making will need to be increasingly mechanized, because humans, no matter how well trained, cannot usefully process data sets of this scale. This may well prove to be the tipping point that sees reliance on algorithms over expert intuition in making clinical decisions become the dominant paradigm. This latter consideration is of great importance. To gain expert intuition, is one of the reasons that health professionals undergo a long period of training. Aristotle called this expert intuition, “phronesis,” or practical wisdom. Phronesis is necessary in clinical decision making because the logic of care is unbounded, non linear and unpredictable and there cannot therefore be an algorithm for every situation. Furthermore, humans reason differently from computers, and take into-account factors that computers cannot (currently) perceive [89]. However as the time spent in medical training has decreased in some countries, arguably the new doctors expert intuition will not be as well developed as their predecessors and clinicians will become more dependent on algorithms. In the developing world where there is a shortage of health professional, dependence on algorithm driven medicine is more likely. This debate of “intuition versus formula” [144] and “expert intuition can we trust it” [119] will continue as technology behind artificial intelligence matures. A midway house may be where the clinical decision-making of the machine, can be immediately understood by the practitioner by

324

Health Web Science, the Semantic Web and Linked Data

providing them with the “arguments” for why the decision was made. This will ensure that the clinician maintains this critical final-arbiter role in the care of their patients. Health Web Science practitioners, must also, demonstrate leadership at local, national, and international levels to drive data liquidity by solving issues related to data sharing and discovery, data interoperability, knowledge representation and exchange, privacy, and web architecture to enable better-connected health information, patients, practitioners, and health researchers thereby improving productivity, efficiency, decreasing health inequality, and utilizing health resources more effectively.

4.3

Workflows and Web Services

The previous sections discussed a variety of technologies and their core uses: Linked Data for capturing the entities and relationships in and between data sets; web Services for exposing both data and analytical tools on the Web in machine-accessible ways; Mashups, which bring together data from many web services and/or Linked Data sources and integrate them into a unified visual representation; and Ontologies, which are used to formalize and harmonize the concepts and relationships in disparate data sets such that they can be merged more easily. In this section, we begin to bring these various technologies together in new and powerful ways that will, over time, let us put much of the burden of data and knowledge discovery and integration onto our machines. Although getting humans “out of the loop” is disconcerting, it seems to be increasingly necessary as both data and knowledge expand. This has led to the expansion of “Workflows” as an area of research in both the pure health sciences, as well as in clinical practice and clinical administration. Workflows are, informally, a description of the flow of information or objects from one location to another in order to fulfill a desired function. For example, in hospital administration “Business Process Management” workflows will describe the process of taking, cataloguing, and storing blood-draws from a patient and billing them for the work; in clinical practice, workflows describe national or international

4.3 Workflows and Web Services

325

clinical guidelines, and are often the technologies behind decision support systems (DSS); in health research, analytical workflows describe, for example, the process of cleaning and processing the data from a microarray experiment. Formal workflows help ensure a process has been done comprehensively, according to a particular standard, and/or help ensure reproducibility, for example, in biomedical research, by acting as the explicit representation of an experiment. Semantics play a variety of roles with respect to workflows. For example, semantics might be applied to DSS as part of the deductive reasoning that helps diagnose a patient based on their symptoms, where the semantic rules encode the possible interpretation of each symptom, and/or the constellation of symptoms that describe any given disease. In biomedical research, semantics are applied to both the data (in the form of biomedical ontologies) such that machines know what any given piece of data “means,” and also to the description of analytical services, such that machines know the purpose of an analytical tool, when to apply it, and on what kind of data. We will now briefly look in slightly more detail at each of these three domains — administrative, decision support, and biomedical research — and explore how workflows and semantics are used to provide solutions within their individual problem-domains. 4.3.1

Workflows in Hospital Administration

Hospitals worldwide are beginning to implement electronic medical records (EMR) as a way to capture, represent, and share all information about patients and their lifetime of care, though there is currently no global standard describing the structure or content of an EMR. EMRs serve two primary purposes — they formalize the way that medically relevant information is represented, and they formalize the way that administrative information is represented. Hospital administrative workflows have more ‘rigidity’ — are more ‘business oriented’ — than clinical guidelines or scientific workflows, with the latter two designed to be “tweaked” by the expert clinician or scientist, respectively. The role of semantics in hospital administrative workflows is largely around the description of various fields in the EMR,

326

Health Web Science, the Semantic Web and Linked Data

as well as the constraints on the data that can be entered into those fields. For example, Health Level 7 (HL7) is the most globally recognized standard for exchange, management, and integration of health care records. HL7 is an expansive project, covering the standards for what information should be in a record, how that information should be organized, and how it should be represented and exchanged over a variety of different protocols (e.g., TCP/IP). Most health-care administrative workflow systems support the HL7 standard; however, this does not necessarily guarantee interoperability between them, nor even between the records they consume/produce. This is, at least in part, due to the reality of hospitals, where each hospital has unique businessprocesses (e.g., the order in which events happen to a record) and unique requirements (e.g., for reporting or billing). This resulting in fragmented or incomplete records, and/or the requirements for data fields that is absent from the existing standards, thus potentially breaking compatibility with that standard. For this reason, formal workflows have not had the impact that might have been expected, given their obvious applicability to this problem [128]. Improving this situation is an area of active research and development, with dedicated scientific workshops around the formal construction, representation, and use of workflows in clinical care (e.g., ProHealth http://mis.hevra. haifa.ac.il/∼morpeleg/events/prohealth KR4HC 2012/). Where HL7 describes the semantics around the structure of a hospital record, the clinical content of that record is also semantically described through a variety of medical coding systems, with arguably the most comprehensive and widespread being SNOMED-CT. SNOMED-CT is a class hierarchy of clinical terms, with ontological features such as “synonymy” — for example, capturing the multiple ways of describing a common clinical phenomenon or disease — and other integrative features such as being multi-lingual. SNOMED-CT includes terms describing a wide range of healthcare phenomenon, including process (what was done to a patient) as well as diseases, symptoms, and causative agents, anatomical sites, and the relationships between them. Thus it is used in both administrative workflows (e.g., billing for a procedure) and for more research-oriented workflows (e.g., selecting/excluding from clinical trials).

4.3 Workflows and Web Services

4.3.2

327

Workflows for Clinical Guidelines and Decision Support

In support of clinical practice, workflows are used to design computerbased clinical consultation systems, either to assist in adherence to clinical guidelines, or to assist in diagnosis. Among the first of these DSS was MYCIN [222], that was designed to diagnose infectious disease, by focusing on bacterimias. Other decision support systems followed, including INTERNIST-I [145], DXplain [18], Isabel [88], Iliad [134], MDX, DiagnosisPro [10], and DiagnosMD [57]. Each DSS differs in its: domain and scope of diagnostic ability; the technology or logics used to achieve the diagnosis; the ability of the system to explain the diagnosis; the set of input data (e.g., symptoms, drugs, history) used; their ability to diagnose in the absence of any given recorded symptom; and their ability to diagnose multiple diseases manifesting simultaneously. The accuracy of these systems ranges between 90% and 96% — the latter approaching the accuracy of clinicians themselves. Recently, an interesting challenge was posed through an attempt to use Google itself as a diagnostic aid [199] revealing that Google searches made the correct diagnosis more than half of the time; however, clearly this does not meet the level of accuracy required for real clinical practice. A recent paper [177] in which semantics are applied to the problem of multi-level diagnosis shows that, in some cases, modern semantically aided DSS workflow systems can achieve an improvement of up to 30% diagnostic accuracy over clinicians, depending on their domain of expertise; effectively the system functions better than a medical generalist, but does not diagnose as accurately as an expert in that particular disorder. This same study showed, interestingly, that the degree of disagreement between physicians was higher than between the DSS and physicians as a whole. Thus, modern DSS systems appear to be quite uniformly reliable, and with high diagnostic accuracy. Achieving accuracy in DSS systems is crucial as we move toward personalized medicine. Individual patient genomes will result in increasingly large and complex single-patient data sets that, given the training and expertise required to understand even single gene polymorphisms are beyond the interpretive power of the clinician. In the (very) near

328

Health Web Science, the Semantic Web and Linked Data

future, however, clinicians will be faced with the full genetic profile of the patient, their current list of medications, the levels of various proteins in their blood (many of which can be affected by diet, exercise, smoking, etc.). Not only would evaluating this data be beyond the capacity of the clinician given the 10–15 min they have for each patient, but also the knowledge required to understand these data is highly specialized and, moreover, constantly changing. It has been noted in other chapters in this monograph that it is now impossible for clinicians to keep-up with the literature, even in their own domain. Workflows, backed by semantics, allow the expert knowledge from thousands of clinicians and clinical researchers worldwide to be formally encoded into analytical strategies that can be selected and invoked, unaided, by the machine, at the appropriate time, under the appropriate circumstances, for the appropriate patient, in much the same way as clinical practice guidelines are currently followed through manual decision-making. There are other contemporary phenomena that will push the adoption of computerized, workflow-driven diagnoses. For example, the availability of Electronic Health Records (EHRs), and their capability to capture the entire medical history of a given patient are increasing. With this much information available, both relevant and irrelevant (or seemingly so), it will no longer be possible to rely on the clinician to know, and interpret this mass of data. Machines will become part of the decision process, with their ability to evaluate all of the data, and sort through relevant/irrelevant information automatically based on encoded and globally shared clinical expertise. In addition, there are social issues around the changing behavior of patients. Patients are becoming much more active in their health-care. One of the consequences of this is the increase in “shopping for care.” As such, it becomes extremely challenging for a clinician to know the current “state” of any given patient, where they are on their treatment trajectory, or even what drugs they may be taking. Again, EHRs, and mechanization of diagnosis within the continuum of care, will help; not only by keeping a record of what treatments the patient currently has, or has had in the past, but also through keeping a record of the intent of the clinician who prescribed this treatment. Based on this

4.3 Workflows and Web Services

329

information other clinicians can judge where, within the best-practice clinical workflow, the patient should be at any given time. 4.3.3

Workflows in Biomedical Research

The role of workflows and semantics as applied to biomedical research is particularly interesting, in part due to the fact that, in this case, the workflow components tend to be more public and open compared to the relatively closed-world hospital administrative systems and “black-box” DSSs. Though Web Services are not exclusive to research workflows, they are (at this time) more prevalent in this domain than either administrative or clinical workflows because the providers of data or analytical tools, worldwide, desire to make these resources publicly available. The generally accepted standard for publication of such resources is the web service paradigm. This, however, makes the assembly of workflows more difficult, since the individual components are not under centralized control, and therefore do not follow consistent standards for the interface, nor the representation of the data. Research workflows are designed to capture, and automate, the flow of biological and clinical data through a process of cleansing, analysis, and even interpretation. Mechanization is important in this case for two reasons — completeness, and reproducibility. It is increasingly clear that researchers cannot keep-up with the literature, even in their domain of expertise [83], and as such, they increasingly rely on distributed data and knowledge resources to ensure that their analyses are fully up-to-date. Moreover, there are increasingly worrisome observations of human limitations with respect to managing and manipulating the large and highly complex data sets that arise from modern biomedical research, and the resulting ease of making errors during data analysis — “the most common errors are simple... the most simple errors are common” [12]. Many researchers lack the skills to programmatically manipulate large data sets, and continue to use inappropriate tools to manage ‘big data.’ Serious errors introduced during data management are difficult to detect by the researcher and, because they go un-recorded, are nearly impossible to trace during peer-review. In addition, the statistical expertise required to correctly

330

Health Web Science, the Semantic Web and Linked Data

analyze high-throughput data is rare, and biological researchers — even those who are extremely competent in rigorously executing the data-generating ‘omics’ experiments — are seldom adequately trained in appropriate statistical analysis of the output. As such, inappropriate approaches, including trial-and-error, may be applied until a “sensible” answer is found. Finally, because manually driven analyses of highthroughput data can be extremely time-consuming and monotonous, researchers will sometimes inappropriately use a hypothesis-driven approach — examining only possibilities that they already believe are likely based on their interpretation of prior biological knowledge, or personal bias toward where they believe the “sensible” answer would be found [15]. Thus, the scientific literature becomes contaminated with errors resulting from “fishing for significance,” from research bias, and even from outright mistakes. These problems are becoming pervasive in omics-scale science. The affordability and accessibility of high-throughput technologies is such that now even small groups and individual laboratories can generate data sets that far exceed their capacity, computationally and statistically, to adequately evaluate. Even more troubling is that it is becoming clear that peer-review is failing to catch analytical errors — whether they are accidental or even willful. While the Baggerly study into high-throughput publication quality [12] triggered retractions and a scientific misconduct investigation [81], the Ioannidis study reveals that, even in the prestigious Nature Genetics, more than half of the peer-reviewed, high-throughput studies cannot be replicated [111]. The failure of peer-review to detect non-reproducible research is, at least in part, because the analytical methodology is not adequately described [111]. In recognition of these limitations, in 2012, the Institute of Medicine published several recommendations relating to proper conduct of high-throughput analyses. These include: rigorously described, rigorously annotated, and rigorously followed data management procedures; “locking down” the computational analysis pipeline once it has been selected; and publishing the workflow of this analytical pipeline in a formal manner, together with the full starting and result data sets. These recommendations help ensure that (a) errors are not introduced through manual data manipulation, (b) there can be no human-

4.3 Workflows and Web Services

331

intervention in the data as it passes through the analytical process, and (c) that third-parties can properly evaluate the data, the analytical methodology, and the result at the time of peer-review. Formal workflow technologies have proven effective at resolving some of these issues [103], and a variety of workflow environments have emerged that aim to support researcher’s in building and using workflows, including Taverna [162], Galaxy [138], Kepler [138], KNIME [203], Pegasus/Wings [82], and SHARE [207]. It is somewhat discouraging, however, that each of these uses its own format for representing the workflow, and are thus largely incompatible with one another. A notable project that is attempting to at least catalogue, if not somewhat homogenize, the research workflow landscape, is myExperiment [85]. myExperiment is a repository for scientific workflows, cast as a social networking website, allowing groups to spontaneously form, and individuals to comment on various workflows, rate them, manipulate them, and share/re-share them. The myExperiment “experiment” is relevant to Web Science not only because it deals with creating a catalogue of research workflows that can be shared and re-used as a way of simplifying and homogenizing the high-throughput research landscape, and sharing individual expertise, but also because of this social aspect where individuals are encouraged to discuss the science around the workflow. Although the myExperiment site encourages workflows to be shared, re-used, and re-purposed, this turns-out to be quite difficult in-practice [217]. Apart from the mutually incompatible workflow representation formats, given that web services are published without any centralized control, they expose highly disparate interfaces that require considerable effort to correctly link-together. Thus, re-using a workflow, even with even minor modifications (e.g., the same analysis on a different organism) can result in extremely complex ‘re-wiring’ of the workflow itself as new services need to be correctly linked to the appropriate input and output data-types. As a result, a variety of projects have emerged that apply semantics to the description of web service interfaces (reviewed in Wilkinson (2011)). The resulting resources are referred-to as Semantic Web services, and these differ from traditional web services in that the operations/functions of the service, and the

332

Health Web Science, the Semantic Web and Linked Data

various elements of the input and output data-types, are annotated with terms defined by ontologies. The intent is that a machine should be able to correctly link-together two disparate services based on its ability to ‘interpret’ the purpose of the service, and also the purpose of each data element. When services are richly semantically annotated, it becomes possible to automatically match output data-types from one service to the respective input data-types of another, even if these datatypes are named differently by the two services. While many Semantic Web service projects have been active for at least a decade, it is difficult to claim that any of them have been notably successful, or widely adopted. As such, the problem of scientific workflow maintenance and re-use remains unsolved.

4.4

Open Research Questions

The Semantic Web is still very new and growing, and there are many research questions and areas that need to be explore, particularly in the health area. One area of concern is in how to determine the source of information encoded in semantic formats — it’s “provenance.” While this is a problem on the Web in general, it is particularly troublesome on the Semantic Web, where data is not organized into visible pages, data is dynamically retrieved and integrated by machines from wide range of sources, and is provided in formats that are difficult for humans to inspect. Provenance metadata, is used to track where data and other information on the Web comes from, including such things as the process by which a piece of data was generated (e.g., database query or algorithmic execution) or the version number of a utilized data set. With this metadata, the quality of data products can be verified, scientific research becomes more reproducible, and researchers who authored the works can be properly attributed [147]. With mashups gaining in number and popularity, and with their reliance on data sets that come from many different sources, we need to have better ways to record, use, and visualize the provenance of linked data to help to ensure trustworthiness and quality. Semantic mashups in recent years have become easier to build and create with the emergence of many off-the-shelf programming tools and

4.4 Open Research Questions

333

libraries, but it is still difficult for non-programmers to do. What new technologies, interfaces, and tools are needed to help data experts who are non-programmers to be able to make better use of mashups and linked data? The linking of the social and Semantic Web has been gaining attention in recent years. There have been some experiments in social semantic applications, but as of yet none have really taken off. What will it mean for average web users to interact with the Semantic Web like they do with the World Wide Web? What tools, workflows, and interfaces are needed to make this more of a reality?

5 Methodologies and Methods for Cross-Disciplinary Study

In a field that is multidisciplinary, each contributing discipline will have its methods of study and evaluation. In an emerging and new discipline, new methods need to be created to address the emergent phenomena once they are observed. There is a belief that eHealth has enormous potential for improving health in all nations. EHealth being defined as an emerging field in the intersection of medical informatics, public health, and business, referring to health services and information delivered or enhanced through the Internet and related technologies. In a broader sense, the term characterizes not only a technical development, but also a state-of-mind, a way of thinking, an attitude, and a commitment for networked, global thinking, to improve health care locally, regionally, and worldwide by using information and communication technology [64]. However Black et al. [31] conclude in their systematic overview on the impact of eHealth on the quality and safety of health care that there is a large gap between the postulated and empirically demonstrated benefits of eHealth technologies. In addition there is a lack of robust research on the risks of implementing these technologies and their cost effectiveness has yet to be demonstrated. The authors of this monograph make the prima facie assumption that 334

5.1 Health Web Science Research

335

the same is true of Health Web Science — where the Web is understood as an intervention. We will therefore review the quantitative and qualitative paradigms as well as the role of the mixed methods including the pragmatist approach in evaluation and argue that as in eHealth [90] all have a place within a precautionary approach philosophy. In this section, we also review the relevant aspects of contributing disciplines and point the direction to where new methods are needed. In the case of Health Web Science, the focus is on health and the challenge is to determine the dynamics of the contributing disciplines and how to measure and adjust the components to achieve the understanding that leads to desired results.

5.1

Health Web Science Research

There are two fundamental approaches to scientific study. The first accepts an absolute truth and therefore an objective reality that can be studied from a detached neutral standpoint. The second argues that there is only subjective truth whereby individuals create, negotiate, and interpret meanings for their actions and for the social situations in which they exist. The first is called the objectivist paradigm, is deductive and infers from the general to the particular or from the population to the individual. The second is the constructivist approach, is inductive and infers from the individual to the population [91]. Modern medical care is influenced by these two paradigms — evidence based medicine (EBM) (objectivist paradigm) and patient centered medicine (constructivist paradigm). They focus on different aspects of medical care and have little in common [24]. EBM tends to be disease orientated rather than patient centered, and also tends to be doctor centered [198]. EBM trials do not necessary reflect the real world and the trials may only involve a small percentage of people in a cohort who meet the inclusion criteria. Patient centered medicine recognizes that the patient is the expert living with the condition and has a contribution to bring to the table. A patient centered approach recognizes that the patient is a source of data and knowledge and that they are an integral part of the development team (co-development) when designing an intervention. In eHealth terms this is a process known as

336

Methodologies and Methods for Cross-Disciplinary Study

apomediation, [67] where the “wisdom of the crowd” complements the wisdom of the experts (health professionals). Social science has made valuable contributions to helping bridge these distinct and opposing frameworks (EBM and patient centered approaches) by focusing on understanding the role of health communication between medical and patient perspectives [196]. Health professional communication has often been found to be inadequate in meeting patients’ needs, for example, where information is insufficient and too difficult to comprehend [139]. To mitigate this, proponents of patient centered care have advocated for both physician training and patient empowerment in order to promote participatory decision making [63] — a process that falls under the umbrella of apomediation [67]. A recent study commissioned by the Commonwealth Fund (studying five dimensions that would be expected from a high performance health system in seven industrialized countries) found that in the UK, patient centered care (defined as care delivered with the patient’s needs and preferences in mind) ranked last overall [48]. This finding suggests that in the UK there is a need to continue the discussion of how to make healthcare delivery more patient centered and perhaps move from the predominance of evidence based medicine influencing our healthcare delivery. Ultimately, for EBM and patient centered medicine to work together, we will need evidence that EBM can be patient-centered and that patient centered medicine can demonstrably improve health outcomes. Some of the challenges related to utilizing methodologically sophisticated research designs while at the same time retaining relevance to the real word of practice and patient care are illustrated when we consider prevalent research designs for evaluating the efficacy of interventions that may benefit patient care. Thus, interventions represent an important research area for evaluating efficacy of new treatment approaches. The main tools of EBM are randomized controlled trials (RCTs) that are the most rigorous way [179] and therefore the “gold standard” in determining whether a cause–effect relation exists between treatment and outcome and also for assessing the cost effectiveness of a treatment [187]. RCTs usually have inclusion and exclusion criteria that are

5.1 Health Web Science Research

337

well defined (good internal validity) and the population under study is homogeneous (i.e., patients do not have other co-morbidities). The question then is, how easily is the evidence from an RCT transferrable to clinical practice (external validity). RCTs may have good internal validity but have poor external validity, i.e., the study is impractical in the real world of medicine. Even if one accepts the RCT as the gold standard “one size does not fit all” and therefore not all interventions need to be evaluated in an RCT [190]. One question that needs to be addressed is “What are the most appropriate evaluations of eHealth interventions and for assessing the health and wellbeing end points under health web science?” As we have already discussed eHealth is patient centered in its philosophy whereas RCTs are not [24]. In addition because people interact with each other, emergence is a property that needs to be studied and is not part of an RCT paradigm. Emergence, in this setting, is the way complex systems and patterns arise out of a multiplicity of relatively simple interactions. The emergent is unlike its components insofar as these are incommensurable, and it cannot be reduced to their sum or their difference [131]. This is a more modern description of the axiom namely “the whole is greater than the sum of the parts” — a phrase which is attributed to Aristotle. Web design, development, engineering, and use need to be understood both at the micro-level, i.e., building and testing web applications, and at the macro-level, i.e., studying the use of the microsystem by many users interacting with one another [107]. That being said, it is important to recognize there would be methodological concerns where we have to reject RCTs completely. Webb et al. [211] conclude that overall there is a small positive effect that Internet health interventions work with an understanding of behavioral change techniques being associated with larger effect sizes. However these studies fall into the silos of EBM. Summarizing the literature Murray [153] concludes that there is a growing awareness that Internet interventions work but it is unclear for whom, for what behaviors and for which medical conditions. She also expresses concerns regarding equity (access to technology and once technology is accessed web literacy, health literacy, perceived behavioral control and environmental constraints).

338

Methodologies and Methods for Cross-Disciplinary Study

Web-delivered interventions have the potential to combine the tailored approach of individual or face-to-face interventions while maintaining the scalability of public health interventions with very low marginal costs per additional user. The main benefits are when used by a large number of people in terms of cost and the potential wisdom of the crowd that emerges from the developing health ecosystems. Researchers are also beginning to research “hot” triggers, and push persuasive technologies to reduce attrition rates [71]. Participatory action research (PAR) is increasingly being utilized as a methodology from a patient centered perspective. This is appropriate in situations where ‘a new approach is to be grafted on to an existing system’ [44]. The aim is ‘to arrive the rules and recommendations for good practice that will tackle a problem or enhance the performance of the organization and individuals through changes to the rules and procedures within which they operate’ [53]. This orientation lends itself well to the agile, iterative approach to software development where each iteration or cycle of development is evaluated and the lessons fed into the next cycle. PAR involves a ‘feedback loop in which initial findings generate possibilities for change which are then implemented and evaluated as a prelude to further investigation’ [52]. Cooperative inquiry comes under the umbrella of PAR. The major idea of cooperative inquiry is to “research ‘with’ rather than ‘on’ people.” It emphasizes that all active participants are fully involved in research decisions as co-researchers. Cooperative inquiry creates a research cycle among four different types of knowledge: propositional knowing (as in contemporary science), practical knowing (the knowledge that comes with actually doing what you propose), experiential knowing, (the feedback we get in real time about our interaction with the larger world), and presentational knowing (the artistic rehearsal process through which we craft new practices). The research process includes these four stages at each cycle with deepening experience and knowledge of the initial proposition, or of new propositions, at every cycle. A third paradigm or framework is therefore emerging in which to approach the scientific study of medicine and which is readily applicable to Health Web Science — Pragmatism [148, 184, 70]. This argues

5.1 Health Web Science Research

339

that medicine is a practice, not just a science or an art. Pragmatism combines qualitative and quantitative methods and argues that clinical reasoning is adductive — “what is best to do for this individual at this time given these circumstances” [89]. Pragmatism acknowledges that “health related actions cannot be explained using rational choice models” and that “human agency is complex, nuanced and subject to a host of influences such as cultural and symbolic meanings, practical constraints and personal priorities.” Greenhalgh [89] reminds us that the ancients had a word for this pragmatic approach “Phronesis” the ability to apply general rules to particular situations. The phronetic social science approach frames its work broadly around four questions: where are we going? Is this development desirable? Who gains and who loses and by which mechanism of power? What if anything we should do about it? Recognition of the needs for mixed methods approaches in this research arena is gaining momentum [133, 46]. Indeed, mixed methods design has become increasingly prevalent for considering the challenges of effective patient care, as well as bottom-up understandings of the role of consumerism in promoting high quality healthcare services. Web-based technology opens the door to patient initiatives including greater patient assertiveness and proactivity toward demanding patient centered care [117]. The discussions above resonate with the conclusions from a completely different discipline, namely a “regreening of Africa project.” Akkerman et al. [4] claim that current mainstream research paradigms for the Web were not able to describe or evaluate their work. They also make the point that the traditional positioning of science “as a one way street, from fundamental research to application is simply not going to work in any real world practice” and that the “phronetic aspect of research that is crucial to empowerment is underdeveloped concluding that web science has yet to make significant steps forward here.” The future of medicine is increasingly becoming P4, i.e., preventative, participatory, personalized, and predictive. Web Science will be integral to understanding these interactions. Griffiths et al. [94] provide a systematic review of the literature on Internet interventions and conclude that “it is important that the researchers clearly state why

340

Methodologies and Methods for Cross-Disciplinary Study

they chose to use the Internet, preferably backing up their decision with theoretical models and exploratory work.” Campbell et al. [39] make suggestions to improve on the Medical Research Council (MRC) framework for the development and evaluation of RCTs using an iterative approach and suggest three outcomes from preparatory work prior to proceeding to RCT. By doing so they try to marry the separate worlds of EBM and patient centered care. These three outcomes are: the intervention does not warrant an RCT; the evidence for the intervention is so strong that it should be implemented; and doubt remains about the effects of the intervention that it needs a trial. We would agree with Murray et al. [155] that in addition the preliminary step is to implement a PAR methodology and then once an intervention is shown to be effective, the challenge is how to integrate it into mainstream healthcare practice (normalization behavior). Proponents of eHealth technology argue that systematic reviews show some improvement and further research is needed to understand better which diseases/conditions, which behavioral modifications should be used and which patient cohorts benefit from the technology and equally as importantly, how all the stakeholders can co-create/codevelop the health applications. However this can be a long drawn out process and health care is in crisis. Appeal therefore in the short term, is made to the precautionary approach that is to keep funding going for web-based initiatives. The “precautionary principle” as applied to eHealth states if an intervention has the potential to cause great harm to a large number of people, then in the absence of evidence of harm, the burden of proof is on those wishing to implement the e-intervention to demonstrate that the danger is not real. The difficulty with this is that it is difficult to prove a negative — i.e., that there is no negative effect. The “Precautionary Approach” takes a softer approach and says lack of full scientific certainty should not be used as a reason for postponing cost effective measures (i.e., the web based-intervention). The perfect storm of demographic change, financial constraint, increase in long-term conditions and current models of healthcare are the drivers to appeal to the precautionary approach while a pragmatic

5.2 Health Behaviors, Incentives and Motivation

341

research paradigm using mixed methods incorporating a phronetic social science approach act as the checks and balances. It is incumbent therefore on those developing health technologies that are being delivered via the web to embrace health Web Science and its aims. Furthermore Hippocrates aphorism, primum non nocere (first do no harm) and Google’s mantra of “do no evil” should be foremost in peoples minds.

5.2

Health Behaviors, Incentives and Motivation

A large part of Health Web Science involves eliciting the participation of the users of a resource to contribute either: (i) their knowledge or expertise (for example, in wikis), (ii) their experiences or opinions (for example in social health sites such as Patients Like Me), or (iii) to simply execute tasks and gather information en masse that are beyond the capabilities of any single individual. The consequence of this reliance on third parties is that the success of a Health Web Science project depends largely on whether the third-party has the incentive to participate. Good (forthcoming 2013) provides an extensive review of the kinds of incentives that are possible in web environments, and what kinds of incentives are most likely to be successful for any given task. He splits these up into explicit and implicit incentives. Among the explicit incentives are money or other tangible rewards, fun, education, and the desire to complete a task of personal importance. Implicit, often “social,” incentives include altruism, doing meaningful work, creating a sense of community, the desire for competition, and the desire for status or reputation. 5.2.1

Incentive and Motivation in Social Communities

One of the promises of the Web for improving the health of users is to find new ways to encourage healthy lifestyles. However, the Web is a comprehensive information resource for patients, patient-advocates, and clinicians may itself not be enough. Furthermore, while web technology can encourage these changes in a patient’s life, in principle, we first need to understand how incentives and motivation work in

342

Methodologies and Methods for Cross-Disciplinary Study

practice. Many disciplines have studied incentivization, i.e., the study of incentive structures, and motivation by developing and testing different models based on both physical and virtual worlds [170]. Research is inconclusive about the connection between motivation in the online and offline world. In the next short section, we cover a few of the more prominent incentive theories, and then finish with some examples of web applications that help encourage health behavior. One of the first classical distinctions in motivation was the split between intrinsic and extrinsic motivations. Self determination theory defined intrinsic as a motivation that comes internally, dealing with self-improvement and challenge seeking [50]. Extrinsic motivation deals with external awards in the form of monitory gain or peer recognition. This broad categorization helps us better understand the source of different types of motivation. Ciolilli [43] breaks up motivation as being ether individual and social in nature. Thus, individual motivation is similar in some respects to intrinsic motivation, but involves efficacy and self-satisfaction. Social motivation involves ideas in wanting to belong to a community, the need to both give and receive support from others, and the desire to help contribute to a community effort. Some Web communities and social networks (particularly in the weight loss and exercise market) have tried to tap into some of social aspects of motivation to help encourage users to exercise and lose weight. Nike+ (http://nikeplus.nike.com/plus/) and Fitocracy (http://www.fitocracy.com) are currently popular examples. These web applications allow users to collect data on exercise and eating habits, and share these results, and compete with both friends and strangers for support, accountability, and sport. The aspect of sport, or gamification i.e., using game playing ideas and props to motivate people to adopt new behaviors or actions) has just started to be used and explored in the area of health. For example, Bayer has created a glucose-measuring device that can connect to a gaming system, helping to encourage children with Type 1 Diabetes to better monitor their blood sugar. “Games are rapidly becoming an important tool for improving health behaviors ranging from healthy lifestyle habits and behavior modification, to self-management of illness and chronic conditions to motivating and

5.2 Health Behaviors, Incentives and Motivation

343

supporting physical activity. Games are also increasingly used to train healthcare professionals in methods for diagnosis, medical procedures, patient monitoring, as well as for responding to epidemics and natural disasters.” Games for Health, a bi-monthly peer-reviewed journal publishes on the development, use, and applications of game technology for improving physical and mental health and wellbeing. It is currently in its second edition. 5.2.2

Incentives: Examples and Experiments

In the preceding section, we discuss incentives and motivations on the Web. In this section, we provide examples. One of the most pervasive examples of a Web Science “experiment” is Wikipedia, and its related initiatives such as Semantic Media Wiki, GeneWiki, DBPedia, and others. The incentives for participation in Wikipedia are multifold, including the desire to improve a global resource for altruistic reasons, the desire to share (or even impose) personal opinion, and the desire to do meaningful work. Given that there are no clear explicit incentives for participation, the extraordinary success of Wikipedia may indicate that social incentives could be, in and of themselves, sufficient to achieve an outcome of enormous magnitude and impact. However, it is not clear that this is reliably replicable. One of the earliest experiments in Social Web Science was executed by Luis von Ahn in his ESP game (2004). The overt objective was to provide amusement for the participants by measuring their “ESP level.” The underlying objective, however, was to “automatically” annotate images on the Web. Users typed-in keywords corresponding to images they were shown, and attempted to type the same keywords as another (unknown) person on the Web, as a measure of their ability to read that person’s mind. The incentive to participate, therefore, was a combination of fun, and of personal desire to somehow be “special.” Another project from von Ahn used a different incentive to overcome the challenging problem of errors in optical character recognition. In ”CAPTCHAs,” unrecognized words from scanned texts are presented to an end-user and the task of the end-user is to type-in the characters that they see in the scanned image. This is used by websites as a

344

Methodologies and Methods for Cross-Disciplinary Study

way to distinguish between humans and machines, for example, to prevent unauthorized access, or to set-up email accounts. The incentive, therefore, is to achieve some desired personal goal, while the rewards are multifold — the Web host protects their resources from undesired automated access, and the work done by the end-user (recognizing and typing the word) improves the quality of digital texts (von Ahn et. al., 2008). There are several examples of incentive-based Web Science projects in the domain of health. One of the most prominent is the Patients Like Me website. The incentive for participants in the Patients Like Me project is multi-fold, including the explicit incentive of helping to become healthier and/or feel better, and more easily keeping-track of one’s personal health, as well as more social and implicit incentives, such as feeling part of a community, sharing advice and burdens, and helping others. Interestingly, as the resource becomes more widely used, and thus richer in its information content, it is becoming an important tool for health researchers and, as a result, the incentive to participate is similarly changing, where participants are now actively adding information specifically with the desire to aid researchers [78]. There are numerous other examples of Web Science incentives, with tasks ranging from genome annotation to predicting protein structures. In each case, there is a carefully planned balance between the cost of participating and the benefit of participating. While our understanding of how incentives work on the Web is still quite shallow, as the number of successful projects grows, the additional examples of success (or failure) will help shape the way we approach the construction of Web Science infrastructures that maximize the rates of participation by the key end-users.

5.3

Engineering Aspects/Medical and Health Applications

Much of the engineering efforts in Health Web Science to date have focused on personalization of the user’s experience on the Web. Early efforts took a “traditional Web” approach, focusing on facilitated search through personal health portals (see, for example, [37]). Semantic technologies were used for search-expansion to discover additional relevant

5.3 Engineering Aspects/Medical and Health Applications

345

pages, map between lay-language and complex medical terminologies to facilitate discovery and better understanding of complex medical literature. A benefit of such portals is that the experts are able to curate the information provided, which (presumably) improves the quality of information presented to the patient; however, it shifts the burden on the curator. For example, nhs24.com (http://www.nhs24.com) provides health information and self-care advice for people in Scotland. NHS Direct in the United Kingdom provides an online Health and Symptom Checkers and Patient Decision Aids (PDAs) (http://www. nhsdirect.nhs.uk/). MedicineNet.com “we bring doctor’s knowledge to you.” WebMD “Better Information. Better health” provides public (http://webmd.com) and private health information, tools for managing health, and support to those who seek information (http:// wbmd.com). MedicineNet.com is owned and operated by WebMD1 ; it provides content a Medical Editorial Board for WebMD. In stark contrast to this approach is the PatientsLikeMe Website. PatientsLikeMe was co-founded in 2004 by three engineers from the Massachusetts Institute of Technology (MIT) for amyotrophic lateral sclerosis (ALS; Lou Gehrig’s disease) a neuro-degenerative disorder. In February 2011, it served 22 chronic conditions and now is open to any condition. PatientsLikeMe is engineered as a social website, with the ability to find individuals with similar symptoms, build ad hoc communities, and engage in discussions. This shows, as was discussed for Wikipedia above, that with the proper incentives and infrastructure, individuals can and will build extremely valuable knowledge bases on their own, with little or no expert curation required. We note, however, that the largely social infrastructure that facilitates peer review and editing may be considered a curation process, however, it is often the case that review and curation is not by experts. From the perspective of Web Science, perhaps the most interesting decision made by PatientsLikeMe is that it is both qualitative and highly quantitative. Patients can rate the severity of their symptoms on a scale, can enter exact dosages of medications, as well as map all of these things over time. This quantitative nature of PatientsLikeMe has resulted in 1 http://answers.webmd.com/organization/40062/medicinenet.

346

Methodologies and Methods for Cross-Disciplinary Study

it becoming a source of high quality and high quantity research data for the medical community, with formal scientific publications based on PatientsLikeMe data arising as early as 2008 (Arnott Smith et al., 2008). It has now been used for research into symptom reporting (Wicks et al., 2012), outcomes research (Frost et al., 2011), and even clinical trials research (Swan 2012). Wireless devices have made automated data collection possible for some kinds of data, such as calories burned throughout the day, steps taken, sweat, and sleep patterns. Devices easily measure these data, however, it is not as easy to measure headaches, brain fog, stomach pain, anxious mood, or many of the other symptom data of the conditions followed on PatientsLikeMe. Notwithstanding, in order to track any of their symptoms, PatientsLikeMe users must input their own data. This was originally accomplished through the Web portal itself, but more recently it is done using an App that communicates with the portal. We anticipate that this will change as an increasing number of consumer electronic devices include wireless capabilities, and are thus potentially able to automate the reporting of health-related data to sites such as PatientsLikeMe, either directly, or via an in-home wirelessready health data collector (for example, recent releases of the Application Developer’s Toolkit for SAMSUNG Smart TVs include modules aimed at wireless collection of medical data from other devices in the patient’s home). However, while some of the symptoms, for example, fatigue may be able to be inferred through quantitative data combined with algorithms and training for an individual patient, it is unlikely that all symptom data will be easy to automatically collect. While the “Website” nature of PatientsLikeMe, and other medically related health portals, provide patients with one-stop-shops for their medical inquiry and socialization, they cannot influence the patient once they browse away from those sites. Given the dubious nature and quality of some of the medical information on the Web, it would be useful to be able to personalize a patient’s entire web experience, as well as provide some information about the reliability or at least the source of additional materials. One early engineering approach to the personalized web experience was recently published (Vandervalk, et al. forthcoming 2013) in the form of a “Personal Health Lens” — a small

5.4 Open and Non-proprietorial Health Web Science

347

javascript button that is placed in a web browser’s toolbar. Upon clicking the button, the Health Lens combines a local database of patient health information, a set of models for medical data interpretation, and the information content of whatever page is currently being displayed. These three resources are passed through a semantic analysis workflow and the page is then dynamically marked-up with patient-specific notes about the various concepts being discussed on that page. For example, the authors used drug–drug and drug–food interactions in their first iteration of this tool, where drug-names in any Web page become highlighted with various warning notes if that drug is contra-indicated with a medicine the patient is currently taking. The authors note that one aspect of the engineering (that was not extensively explored) was that the models that guide the interpretation of the data could, themselves, be personalized — for example, by the patient’s clinician — as a means by which the clinician could “look over the shoulder” of their patients, and guide their self-exploration of medical information on the Web in the context of the Clinician’s plan for them.

5.4

Open and Non-proprietorial Health Web Science

Current models of health care delivery throughout the world are unsustainable and if underlying problems are ignored “they will overwhelm health systems, creating massive financial burdens for individual countries and devastating health problems for the individuals who live in them” [169]. The World Health Organization (WHO) is the United Nation’s organ with a mandate in global public health and recognized in 2005 “that the use of information and communication technologies (ICT) for health, or ‘eHealth,’ represents one of the key instruments for health care delivery and public health.” This was then ratified by the World Health Assembly in its resolution on eHealth in the same year eHealth is a term which has only been around since the start of the millennium and can be loosely described as any “health service and information delivered or enhanced through the Internet and related technologies. Importantly it is also a state-of-mind, a way of thinking, an attitude, and a commitment for networked, global thinking, to improve health care locally, regionally, and worldwide by using

348

Methodologies and Methods for Cross-Disciplinary Study

information and communication technology [64]. Interestingly in their review of the definition of Health 2.0/Medicine 2.0, van De Belt et al. [22] commented that “apparently government is not yet an active party in the development of Health 2.0/Medicine 2.0.” Related to eHealth are the terms telecare, telehealth, telemedicine, and telehealthcare (see Appendix). There is a growing awareness that Internet interventions work, but it is unclear for whom, for what behaviors and for which medical conditions. Furthermore, there are also concerns regarding equity (access to technology and once technology is accessed web literacy, health literacy, perceived behavioral control and environmental constraints). 5.4.1

We are Now Living in the “Anthropocene Age”

The “anthropocene age” — a new name for a new geologic epoch — one defined by our own massive impact on the planet [124] where climate change [39, 82] and degraded environments are now being recognized as additional factors having an impact on health and where action is urgently needed to mitigate against their effects [105]. As patients’ conditions become increasingly complex and monitoring of environments through sensors more ubiquitous, patients, their advocates, and their health professionals need to be able to integrate more health data about them and health care systems — the web of linked data [30] becomes an increasing necessity with decision making algorithms commonplace. Long term chronic conditions (LTC) or noncommunicable diseases (NCD) bring about a long period of poor health and diminished function impacting on quality of life and health care costs. LTCs are the “leading cause of mortality in the world and are an invisible epidemic that hinders the economic development of many countries [212] and must be addressed [202]. Professor Ralph Keeney (a decision analyst at Duke University’s Fuqua School of Business) argues that 55% of deaths are caused by poor choices secondary to people being ignorant and are therefore potentially preventable [122]. LTC/NCDs secondary to smoking, overeating, and unsafe sex for example can arguably be reduced through education and behavioral change. Delivery of healthcare using

5.4 Open and Non-proprietorial Health Web Science

349

the Internet may be able to mitigate these poor choices. For example, social media and viral up scaling has found traction in business and industry and community powered healthcare taps into what research from the Pew Internet and American Life Project calls the mobile [112] and the diagnoses differences [74] Horrigan claims that the mobile difference is that people who have Smartphones are more likely to share and create — thus they are participating instead of simply consuming. The further claim is that the diagnoses difference is realized, holding all other demographic characteristics constant, that people with a chronic disease diagnosis significantly increases an Internet user’s likelihood to say they both contribute and consume user-generated content related to health. Thus, these “diagnosed individuals” are learning from each other, not just from institutions. There is no such thing as a neutral education process [76] and health professionals must accept and embrace the role of nudging people to make beneficial health care decisions under the umbrella of libertarian paternalism. Libertarian paternalism in health care means being free to make choices, while being nudged away from the worst ones (as decided by the health professional) by the way in which the choices are presented. The science underpinning this involves behavioral psychology [211], and study of engagement of the health tools by the public. Thus, there is an urgent need to evaluate new Internet and related technologies that reduce disease burden, keep patients at home, maximize health professional time, and encourage and support self-care. In the UK, the Delivering Living Lifestyles at Scale project aims to show how assisted living technologies and services can be used to promote wellbeing, and provide top quality health and care, enabling people to live independently — including a preventative approach. The recent reporting of the Whole System Demonstrator Programme (which is the largest based randomized control trial of telehealth and telecare in the world from England) showed the potential of telehealth/telecare technology to reduce mortality, reduce the need for admissions to hospital, lower the number of bed days spent in hospital and reduce the time spent in Accident and Emergency [55, 195]. However as already noted, engagement with public, professionals and patients is mandatory in the design of these services otherwise, “the

350

Methodologies and Methods for Cross-Disciplinary Study

risk that the service will be abandoned or not adopted at all is substantial.” These were the conclusions drawn from the English HealthSpace experience [92], which had been preceded by “policy documents” published in 2005–2008 written in a hopeful upbeat style” trumpeting the eHealth agenda. HealthSpace was a free, secure online personal health organizer provided by the UK National Health Service (NHS) that can help people manage their health, store important health information securely, and find out about NHS services near them. The problems implementing HealthSpace and similar projects not only in the UK [69] but also in other parts of the world [135, 121], has been to see the healthcare IT industry as overpromising and under delivering, destroying trust, and causing skepticism [192]. Robertson et al. [175] concluded “If a single lesson is drawn to be drawn from the NHS IT project then, it is that high-level political leadership should create the framework to kick-start IT-based reform, but that thereafter it should be professionally led with a realistic timescale, recognizing that experience with new systems, conflicting interests, and changing circumstances will require careful negotiations along the way.” It is increasing being recognized that the new world of healthcare is more “patient centric” and about partnership with a flattening of traditional top down hierarchical approaches as the Salzburg statement to shared decision making attests whereby patients and clinicians are called on to work together to be coproducers of health [61]. Co-production and co-creation are recognized principles within the NHS and recognize that those who are affected by a service “are best placed to help design it” i.e., a bottom-up approach is employed [158]. In addition, there is increasing recognition that different methodologies need also to be employed to deliver IT health care solutions and evaluate them [92]. In this paradigm, incentivization is the carrot and stick approach is an idiom that refers to a policy of offering a combination of rewards and punishment to induce behavior. Positive results using the stick (of the carrot and stick approach) have been achieved recently related to smoking (e.g., taxation and legislation on tobacco where in England results show benefits for health, changes in attitudes and behavior and no clear adverse impact on the hospitality industry [19]. Other countries as well are exploring intervention strategies such as smoking bans

5.4 Open and Non-proprietorial Health Web Science

351

(in workplaces, bars, and restaurants), labeling requirements describing potential negative impacts of smoking, as well as taxing and pricing strategies, and some analyses show statistically significant correlations between some of these strategies and decreased smoking prevalence [141]. However, this approach can be viewed as the intrusion of the “nanny state” where government policies are overprotective or interfering unduly with personal choice. On the other hand, the introduction of a 2.3% saturated fat tax Denmark in 2011 was scrapped the following year because of unintended emergent properties [60]. Life coaching [180] and health coaching services have been proposed/emerged as a means to improve lifestyles and impact on the prevalence of NCD/LCD. These services are available in many formats, e.g., personal health coaches in gyms, peer groups like Weight Watchers, and most recently as “Apps” on smart phones (e.g., RunKeeper). Service companies have emerged that offer health coaching programs both to individuals and — increasingly — to corporations that want to invest in the health and wellbeing of their employees. And then there are the “SelfQuantifiers,” individuals who record certain aspects of their lives in great detail, analyze this data, and conduct self-experiments in the interests of improving their lives (http://quantifiedself.com/about/). All these strategies aim at changing health behaviors primarily through feedback and coaching. Advanced sensors, mobile phones, and social media are the core technologies deployed. Professional health coaching services have significantly evolved over the past years (e.g., HealthRageous, www.healthrageous.com) and have consolidated into a three-tiered service: face-to-face consultations, virtual contacts (by email, telephone, social media) and automated services. The business model is based on trying to provide automated services that support users in managing their health condition and only use the upper layers when goals are not being met. These services have been shown to be somewhat effective but are challenged in demonstrating sustainable business models. These can be viewed as the first wave in attempting to engage people to be responsible for their health and wellbeing using tools and services that give them the ability to be responsible (response-ability). These solutions have a high degree of failure with time, as they are

352

Methodologies and Methods for Cross-Disciplinary Study

not supportive enough and the initial impulse often backfires as people stop using them. An additional issue is that the solutions are often not based on science or medical evidence [1]. We need more effective actions that complement the current ones, extend their reach, and provide more support to individuals in a more sustainable way. 5.4.2

Self-management

Supporting individuals in non-communicable disease (NCD) prevention is a complex multifaceted issue that, as so, has been thoroughly studied and analyzed. In secondary prevention, the Chronic Care Model (www.improvingchroniccare.org) is a well-established clinical care model centered on the patient aiming to support the individual in managing his/her disease. The basic idea is that ultimately the individual is responsible for managing their chronic condition (disease) and others can only support and empower him to act in this proactive role. There are several variations to what Self-Management means and requires from individuals, e.g., the Seven Principles for Self-Management (Flinders Program http://www.flinders.edu.au/ medicine/sites/fhbhru/self-management.cfm#sixPrinciples) that highlight the need for individual responsibility to be accompanied by ‘response-ability’ that provides the capacities and tools to maintain the individual’s efforts: 1. Have knowledge of their condition. 2. Follow a treatment plan (care plan) agreed with their health professionals. 3. Actively share in decision making with health professionals. 4. Monitor and manage the signs and symptoms of their condition. 5. Manage the impact of the condition on their physical, emotional, and social life. 6. Adopt lifestyles that promote health. 7. Have confidence to access support services and the ability to use them.

5.4 Open and Non-proprietorial Health Web Science

353

These concepts emphasize that the central resource in managing health and lifestyle and in preventing NCDs is the individual. Having the individual as the chief actor and supporting him are essential requirements for an effective prevention strategy. The European SSA PREVE (Specific Support Action collaboration identifying research directions for disease PREVEntion) studied this issue in detail, while analyzing the research opportunities for Information and Communication Technology (ICT) in this domain. The PREVE White Paper (http://www.preve-eu.org/) summarized the project’s main findings and included the key principles for creating personalized support systems for behavior change: • People cannot be forced or manipulated to make a lifestyle change. Change must be based on free will. • In order for a lifestyle change to take place people must have sufficient awareness, motivation and resources, and live in an environment that does not hinder healthy behaviors. • A personal profile can be constructed that comprises the individual’s clinical risk factors and other biomarkers, health behaviors, values, preferences, abilities, intervention goals, demographics, etc. The profile is as dynamic as the person and the social context evolves with time. Even for similar personal profiles, the optimal way to reduce or overcome health risks presents different faces depending on moments of life, different situations or events, present and past. Therefore, interventions need to be tailored to match the personal profile of a person. People value health differently. On the average, health is taken for granted and people are not interested in investing in health if it conflicts with more immediate short-term benefits. Only when health is lost people become interested in regaining their health. Therefore, it may be necessary to promote short-term gains such as general wellbeing and positive experiences (entertainment, games, fun, enjoyment, social networking, etc.) and leave the health benefits to the background as additional benefits. The key observation from PREVE is that individual behavior is the result of interactions between physical, psychological, and social

354

Methodologies and Methods for Cross-Disciplinary Study

processes. It always takes place in a specific context that influences the available options and resources for healthy behaviors. Buy-in from the youngest digital citizen to the oldest with a “health-e-wealth” incentivization plan would be the goal within a health-e-community. 5.4.3

Examples of eHealth Self-monitoring and Guidance Tools

Professional health coaching and life coaching have been available on the Internet for more than a decade in neuropsychiatry, general medicine, and special education. Some coaching-related apps were described above. This section provides examples of how e-diaries have been implemented successfully both as interventions and as support for non-electronic interventions. Intervention recording is a procedure used to promote positive personal change. Supportive recording is a procedure that enhances the impact of another form of intervention such as psychotherapy, medication, or special education. Supportive recording provides feedback that can guide decisions about modulating, reconfiguring, or redirecting the primary intervention for maximum benefit. Two applications are described in this section: (1) web and mobile technology for guiding students with Asperger Syndrome in school and (2) web and mobile technology for monitoring and management of headaches, fatigue, and cognitive lapses with adults with neurological disorders. The programs described involved both types of procedures. These proactive intervention and supportive recording paradigms engaged participants in their care and fostered strong patient–clinician collaboration. Asperger Syndrome (AS) is a developmental disorder. It is an autism spectrum disorder (ASD), characterized by a greater or lesser degree of impairment in language and communication skills, as well as repetitive or restrictive patterns of thought and behavior. Individuals with AS have normal to very high intelligence, yet frequently have challenges with “executive functioning,” such as, organizing, initiating, analyzing, prioritizing, and completing tasks. They have difficulty knowing what to say or how to behave in social situations. Individuals with ASD often have tremendous interest and abilities in visually represented information such as technology, computers, and

5.4 Open and Non-proprietorial Health Web Science

355

video games. In partnership with Dr. Gary Mesibov from Division TEACCH at the University of North Carolina/Chapel Hill, SymTrend (www.symtrend.com) developed tools and a paradigm to take advantage of these interests and abilities to improve social pragmatics, selfregulation, and executive functioning. Mesibov and SymTrend completed two NIMH-funded studies of this paradigm: first with seven teens with AS in a school for students with ASD and then with approximately 50 students with ASD in public middle and high schools in Massachusetts and North Carolina [130]. The protocol was used daily in two classes. Students and professional observers recorded the students’ behavior, mood, energy, and organization on PDAs (initially Palm, then iPod Touch). Students were taught in class to self-monitor their positive and negative behaviors.

c The Science of Me , an approach and curriculum for individuals with ASD developed at Aspire (www.mghAspire.org) extended the above paradigm in a summer camp. Their curriculum focuses on stress management and social emotional intelligence (SEI) and uses several forms of technology. In that program, the SymTrend software facilitates multifaceted behavioral observation, coaching, team-communication, and graphical data to form a treatment support system for use by professionals and individuals diagnosed with an ASD, behavior challenges, or other neuropsychiatric issues. These objectives are achieved through three critical components: (1) recording: simultaneous mobile self- and observer-reporting about behaviors, social interactions, emotional responses, and internal physical states (or inferred states by the observer), (2) guiding: context-sensitive reminders, guidance, and education delivered in a timely fashion on the mobile device, and

356

Methodologies and Methods for Cross-Disciplinary Study

(3) interpreting: web-based charting of progress over time, data analysis, and discussing results from varying perspectives. Data collection screens relevant to behaviors/emotions/internal responses of major interest are included in combination with checklists of individual-specific potential triggers/stressors, and contextual information. Textual or graphical guidance screens can incorporate treatment “homework” or general education material designed for the specific individual or derived from the treatment methodology (e.g., CBT negative thinking exercises). The questions ranged from basic health information (e.g., “Did you sleep through the night?”) to specific information (e.g., “Check any stressors you thought about during the previous activity”). They were reminded to note the physiological changes that accompanied their reactions to stress (e.g., discomfort in their stomach or chest). Interventions used were also noted (“Check any relaxation strategies you used. . . ”). Hindrances were included such as “During the last activity ” (e.g., looping thought, having my own way, etc.). I was stuck on During moments of stress, frustration, and cognitive inflexibility, they were reminded of what they were learning and encouraged to implement positive thinking self-talk, to think socially, and to use relaxation techniques among others. By incorporating other curriculum vocabc they are able to track the usage of multiple ularies into SymTrend curricula strategies across the day in various settings and under various conditions. Learning takes place through individual and small group instruction at least weekly, with a frequency that varies with the setting. Through review of the visual representation of data, individuals and staff are able to discuss “personal data” in an objective manner. In this way, the “negative feedback” is not as upsetting or emotionally triggering, allowing the participants to develop greater self-awareness and better SEI. The ultimate goal is to teach the teens how their behavior impacts themselves and others. For instance, how exercise or lack of exercise or sleep may impact their participation or attention during the day; how their individual participation impacts the overall group performance; or how their reaction to a stressful trigger can either help them “regroup or explode further.” [137, 136]. The graphic representation enables the

5.4 Open and Non-proprietorial Health Web Science

357

students to explore their feelings and gain insight into their reactions to others. Using Web and Mobile Technology to Manage Fatigue, Headaches, and Cognitive Lapses. A psychologist and two stroke neurologists at Massachusetts General Hospital in Boston developed a web and mobile technology as a prophylactic approach to manage patients [38, 129]. The cases involve common symptoms that are at times neurologically ambiguous: headaches, fatigue, sleep disturbance, somatosensory spells, social impropriety (including emotional distress), and cognitive lapses. These patients often cannot report whether circumstances, or what circumstances may have influenced symptom occurrence or intensity. These patients come to MGH when their condition has not been satisfactorily resolved in their local community setting. The MGH Neurology technology is an extension of the neuropsychological evaluation and involves a patient-specific sampling of the patient’s symptoms, feelings, and event stresses for at least three weeks (they have followed patients for as long as four years). These clinicians hypothesized that reports of symptoms and ongoing activity near the time of occurrence could be reliably captured and quantified with an accuracy that exceeded what patients could report later in a visit. Patient recall at clinic is undoubtedly compromised by limitations of memory capacity. It is, however, additionally likely that patients are too distracted to fully appreciate the circumstances of symptom expression or these circumstances are too elusive to be grasped because of the complexity of the event–symptom contingencies involved. The clinicians further hypothesized that a small number of situational parameters would have the greatest influence on symptom expression and posited that the correlation between these parameters and symptom expression would prove diagnostically informative or treatment relevant. The objective of the evaluation and treatment is to identify and then manage the event stresses that trigger the spells. The patient makes entries into an Internet-based diary three times/day. The entries capture a spectrum of information; there are (1) eight generic questions about feelings and reactions to stress, (2) three questions

358

Methodologies and Methods for Cross-Disciplinary Study

regarding sleep, (3) several patient specific questions about symptom expression, and (4) several patient specific questions regarding forms of event stress that occurred that day (e.g., tasks, psychosocial, financial, marital, childcare, etc.). The electronic diary software (symtrend.com) (1) graphs these entries over time and (2) correlates symptom expression with stress, sleep, day of the week, etc. These correlations enhance symptom management either (1) through the clarification of the diagnosis these correlations afford, or (2) through more detailed understanding of how activity impacted symptom expression. Ongoing therapy, when continued, involves general health symptom management, sleep hygiene education, exercise, stress/anger management activities, reminders to “take stock,” and periodic reviews of charts of the symptom course with a clinician. The MGH group documented the utility of the e-tracking in 37 cases with a success rate near 66% [129]. 5.4.4

Web-mobile Individual-professional Collaborations

The above examples of e-diary use in diagnosis and education/ intervention demonstrate how Health Web Science is emerging to benefit patients and professionals. In these cases, patients used web/mobile technology in conjunction with treating clinicians and coaches. Emergent technologies are being designed to minimize unnecessary face-toface contact with health services. In these paradigms, however, the collaboration of the individual and professional enabled through the technology was critical. While the emergent technologies will reduce the cost of care by reducing the frequency and duration of face-to-face contact, effective technologies such as these often require at least some individual–clinician interaction for feedback and interpretation of findings [205, 157].

6 Health Web Science and Protection of Privacy

“The right to privacy, in most interpretations, refers to individual control over personal information.1 It is the ability to restrict access to the information, how the information is held, and what can be done with it. Privacy law refers to the laws that deal with the regulation of information about an individual, regardless of where the information was generated or collected. Privacy laws also regulate storage and use of these data. Privacy is a complex area because privacy laws are considered in the context of an individual’s privacy rights or reasonable expectation of privacy and these vary in different countries, cultures, and situations. In this section, we introduce the basic concepts of privacy-specific issues that are important to be aware of in web-based activity related to health, using examples from the United States and Canada, It is apparent the need for privacy analysis is in the developing world, however, the bulk of research has been done in the developed world. Protection of an individual’s right to privacy in medicine is particularly challenging because the practice of medicine is often personal, but 1 From

a U.S. perspective, the right to privacy is actually more of an autonomy from the

State.

359

360

Health Web Science and Protection of Privacy

medical knowledge is also valuable when it is generalized. Today’s complex conditions often require that personal health information (PHI) be shared in order to provide optimal care. But this risks diminishing the patient’s control over their own data by making it available to others. Controlling who has access to personal medical information is important because of the potential for other individuals and organizations to abuse the information disclosed. At the same time, research into new medical treatments and interventions can be hampered by a lack of access to personal medical data. This tension between individual rights and collective benefits frequently lies at the heart of privacy debates in healthcare.

6.1

Protection of Private Health Data on the Web

Private spaces, both real and virtual, are by definition separate from the rest of society. They are subject to restrictions regarding who can access them. The word “PRIVATE” painted on a door in a public building means “keep out” -access is restricted to only those authorized to enter. There are times however, when the restriction is not indicated directly, nor is it obvious, and as a result, we may trespass unknowingly. A nice stroll in an afternoon could be fatal if the trespassing is into the wrong place. The term trespassing is typically seen with respect to land access, the terms breach, infringement, and violation are more common in the context of law. Interpretation of the law grants individuals privacy rights. In health, the rights refer to personal information or personally identifiable information, the protections afforded by these rights, such as how the information is managed, and the consequence or action that must be taken when a violation occurs. The protection of private health data on the Web therefore reflects concepts of individual privacy that arose from views of solitude, confidentiality, and personal physical space dating back to the early philosophers. Modern Western privacy law dates back to scholars like Prosser, Warren, and Brandeis. Thus, principles derive from case law interpreting the United States Constitution and the Bill of Rights, in particular the Fourth Amendment. In Griswold v Connecticut [381 U.S. 479 [1965], a seminal case that determined that a state cannot

6.1 Protection of Private Health Data on the Web

361

enact a criminal law banning the sale of contraceptives as such a law violated the privacy married couples are entitled to, the United States Supreme Court expanded on the right of privacy described in Boyd v. United States, 116 U.S. 616 as protection against all governmental invasions “of the sanctity of a man’s home and the privacies of life.” The court referred, in Mapp v. Ohio, 367 U.S. 643, 656, to the Fourth Amendment as creating a “right to privacy, no less important than any other right carefully and particularly reserved to the people,” and in Griswold, the court included “the marriage relationship” in the zone of privacy protected.2 Privacy rights are included in Section 8 of the Canadian Charter of Rights and Freedoms. Clearly a multifaceted concept to begin with, information technology has further complicated the interface between individual privacy rights and governmental regulation seeking expediency as regards the sharing of information in the digital age. The architecture of the Internet, a vast and complex array of interconnected networks across the world, set out functional requirements for the collection, use, and disclosure of information.3 The Internet was intended to be a place to traffic information — lots of it — generated by users about unlimited topics (including themselves), generated by machines about other machines (e.g., protocols), and generated by machines about users. A lot of this information we now know to be identifiable to the individual to whom it refers, and is often word-searchable. The architecture of the Internet in fact encourages us to upload and transmit a lot of Information for use at a later date or by others. With this architecture arise three core functional concerns that raise privacy issues. First, unlike people who can forget, and paper filing systems that can be discarded or destroyed, the Internet does not forget. Second, identity on the Internet is generally unconfirmed, both for author and subject: i.e., in order to achieve authenticity, it must be built into the system using nonrepudiation technology in order to 2 See

Beaney, The Constitutional Right to Privacy, 1962 Sup. Ct. Rev. 212; Griswold, The Right to be Let Alone, 55 Nw. U. L. Rev. 216 (1960). 3 See, for example, the Requests for Comments (RFCs) from the 1970s detailing the functional specifications.

362

Health Web Science and Protection of Privacy

ensure the sender is the sender, the receiver is intended receiver, and the data has not been tampered with during the sending process — critical for integrity of diagnostic processes. And, third, the Internet is an excellent tool to aggregate disparate fragments of information that enables compilation of this information to identify, examine, and profile individuals. 6.1.1

Exposure of Medical Data and the Permanence of Disclosure

While it is certainly the case that patient files were misplaced, lost, and stolen prior to the time when electronic systems were used in healthcare, there were no mandatory requirements for reporting these breaches. With the passage of legislation guiding the development and implementation of online electronic health care applications and records, reporting requirements for breaches have become mandatory. They are logged and publically available. In the US, the Department of Health & Human Services tracks breaches and makes them available for download on their website4 and the Privacy Rights Clearinghouse5 tracks breaches. Records compiled by the Department of Health and Human Services reveal that personal medical data for more than 11 million people have been improperly exposed during the past two years alone. Since passage of the federal stimulus package, which includes provisions requiring prompt public reporting of breaches, the government has received notice of 306 cases from September 2009 to June 2011 that affected at least 500 people apiece. A recent report to Congress tallied 30,000 smaller breaches from September 2009 to December 2010, affecting more than 72,000 people. The 4 U.S.

Department of Health & Human Services. Breaches affecting 500 or more individuals: Available from: http://www.hhs.gov/ocr/privacy/hipaa/administrative/ breachnotificationrule/breachtool.html. 5 Privacy Rights Clearinghouse. Chronology of data breaches security breaches 2005 — Present. Updated Date: April 12, 2013. Available from: http://www.privacyrights.org/ data-breach.

6.1 Protection of Private Health Data on the Web

363

major breaches — a disconcerting log of stolen laptops, hacked networks, unencrypted records, misdirected mailings, missing files and wayward e-mails — took place in 44 states. Regardless of reason or intent, breaches happen. Privacy breaches in eHealth are increasing as the use of electronic portable media is used; consider the ease with which an authorized user can copy electronic records to a laptop, external hard drive, USB stick, mobile phone, or tablet. Computer systems are challenged when it comes to human error, the ability to create work-around (such as sharing a password) and in some cases malicious intent. Once personal health information is accessed in an online system, it is also much easier to transmit, disclose, and retain, whether intentional or not. One way to protect privacy in this context is to block out identifying information, but leave the remaining record intact. Electronic redaction accomplishes this by blocking, masking, blurring the identifying information. For example, this is an example of part of a form with personal information: Name: Joanne Luciano Date of Birth: September 13, 2013 Patient ID: 1234567890 While this is an example of the same form with the personal information redacted: Name: Date of Birth: Patient ID: However, once information is released online, even accidentally, any meaningful redaction is extremely difficult. Online information is captured in duplicate form in multiple places, such as cached web pages and to logs, by permanent tracking mechanisms. In short, any privacy breach that involves PHI online cannot be undone. In a paper world, a record can be left open for a passer-by to glimpse at, or a file folder stolen. Online, once a medical record is released, it is permanently available and privacy assurances are meaningless. The permanency of

364

Health Web Science and Protection of Privacy

error in an online environment combined with PHI as content makes decisions on disclosure permanent. 6.1.2

Identity, Authentication, and Authorization

The concept of identity is used to deal with the administrative functions in online systems that control access to PHI. Identity on the Internet is generally unconfirmed, both for author and subject. Until the identity of both the sender and the receiver of the data are confirmed, we are uncertain as to its authenticity. The concepts of authentication and authorization are relatively familiar in electronic environments. Where authentication refers to identifying an individual using an application, for example, by username and password; authorization is about what specific information the authenticated user has access to. Despite distinct identities, patient privacy may not be guaranteed. Issues of password sharing in healthcare are especially problematic; audit trails are not reliable and clinicians are used to having administrative assistance for patient information management. Misdirected emails are another method by which information can be easily misdirected online. In Canada, several enforcement orders under the health privacy legislation in the biggest province (Ontario) have been related to the mismanagement of eHealth records by researchers. The ease with which authorized users may access a large number of records from core electronic health records systems has resulted in situations where PHI is lost or stolen on unencrypted electronic devices. Why are authentication and authorization important in this context? Proof of data integrity is especially critical in an eHealth environment. Healthcare providers need to make time sensitive decisions on the provision of care, particularly in emergency situations that will depend on the accuracy and integrity of the information available to them on the network. These decisions will be based on presenting symptoms, but also treatment and (in particular) drug records provided in eHealth software and databases at the same time. In cases where information provided online is inaccurate, not up-to-date, or (worst case scenario) deliberately manipulated, the patient’s life is at risk. The importance of data integrity to patient health raises significant questions surrounding the accountability of those responsible for the

6.1 Protection of Private Health Data on the Web

365

data. Repudiation means to reject the truth or validity of something. Non-repudiation refers to a situation where the maker of a statement is not able to deny the validity of the statement or contract. In a legal setting, it refers to the authenticity of a signature being challenged or “repudiated.” For example, if someone’s signature is on a registered mail receipt, they cannot deny that the letter was received. A digital signature6 is an electronic signature that can be used to authenticate the identity of the sender of a message or the signer of a document. A digital signature alone may not always guarantee nonrepudiation. Other approaches are also used, such as capturing unique biometric information and other data about the sender or signer that taken together would be difficult to repudiate (Rouse 2008). ISO/IEC 13888-2:2010 provides descriptions of generic structures that can be used for non-repudiation services, and of some specific communicationrelated mechanisms which can be used to provide non-repudiation of origin (NRO) and non-repudiation of delivery (NRD). The American National Standards Institute (ANSI) sponsors the Healthcare Information Technology Standards Panel, which includes a number of specifications available online in respect of transmitting health information in electronic (and online) systems. Of these, HITSP/C26 Nonrepudiation of Origin sets out the requirements for proof of the integrity and origin of health documents. The idea of ‘identity theft’ can also be extended to online systems containing PHI. Identity theft is an old crime; but medical identity theft has consequences that go beyond those of traditional identity theft. In an eHealth environment, the accuracy of medical records is crucial. Similar to nonrepudiation concerns, treatment decisions will be made based on the eHealth records for each patient. So will insurance and coverage decisions. If Alice uses Bob’s information to receive treatment, the electronic record becomes part of Bob’s file. That may include a different blood type, test results, and/or diagnosis that do not apply to Bob. This information can be used to make decisions on everything from treatment to employability, depending on access to these eHealth 6 http://searchsecurity.techtarget.com/definition/digital-signature.

366

Health Web Science and Protection of Privacy

records. Authorized users include: insurance companies and government agencies, and specifically in the United States also the Medical Information Bureau (some Canadian data as well), IntelliScript and MedPoint (prescription drug purchase histories), employers, courts (under subpoena), agencies (e.g., Centers for Disease Control), and marketing companies (as a result of participating in informal health screenings). 6.1.3

Data Aggregation and the Risk of Identification

The platform of the Internet poses specific risks to individuals’ privacy that can have real world implications ranging from embarrassment about personal health information that they did want made publicly available (or specifically available) to limiting insurance coverage and/or employment opportunities. Health information, including demographic, employer, past, present, or future physical or mental health or condition of an individual, payment, provision, or other data relating to an individual could be used to identify an individual and is therefore protected. However, the Internet has made it possible to combine enormous amounts of data that independently are not considered to be identifiable, so that when these data are merged, it may be possible to identify and individual. Indeed, this was demonstrated to be the case using personal genomes, surname inference, and public data sets, all available on the Web [97]. Furthermore, when click-stream data is combined with personally identifiable data that users type in when filling out online forms, web surfers may be profiled in ways that raise serious privacy concerns. Imagine, for example, if employers started inferring health information about their employees (or prospective employees) based on information about visitors to medical- or health-related web sites. Despite these risks, data aggregation can have collective benefits. Secondary use is a privacy concept that describes when information is used for a purpose other than that it was originally collected for. When a provider collects health information, it is for the purposes of providing care. Any other use, such as public health surveillance, research, health management and planning, insurance, and so on, is considered a secondary use of that information, which may or may not be known

6.1 Protection of Private Health Data on the Web

367

to the patient. Online medical records are especially vulnerable to secondary use because of ease of access for authorized users, while the use itself may or may not be authorized. Data aggregation therefore raises difficult questions about how to obtain the benefits from activities such as public health surveillance or medical research while protecting the privacy of patients. One example of secondary use is research. While covered by health privacy legislation in some places, for example Ontario’s Personal Health Information Protection Act (PHIPA), the enforcement of the requirements is often left to Research Ethics Boards (REBs or IRBs in the United States) comprised of academics and community representatives reviewing research proposals once submitted. The threat to privacy of research that is published is closely linked to the appropriateness of the de-identification protocols applied by the principal researcher. De-identification is the removal of identifying information such as the patient’s name, medical record number, birth date, Social Security number, address, etc. from medical records, in order to protect patient privacy. The use of PHI for research purposes is closely linked to the privacy issues of data aggregation. In most health privacy legislation, research is a permitted non-consensual use of PHI by authorized healthcare providers (e.g., clinicians). In practice, research teams usually include students and other non-employees of an organization that holds eHealth records (e.g., hospitals, labs, etc.). More importantly, the concept of de-identification for research purposes has been difficult to achieve particularly when so much information is published in databases accessible on the Web. In 2006, Mailin evaluated the IdentiFamily tool on real world data for a state’s capital city and demonstrate unique identifiability for approximately 70% of the population. IdentiFamily is a tool to evaluate the anonymity of pedigrees prior to disclosure and intended to facilitate the design of formal privacy protection techniques. “To have the illusion you can fully protect privacy or make data anonymous is no longer a sustainable position” [141]. In the next section, we discuss the legal implications of health on the Web. Along with the increasing utility of the Web as a resource for

368

Health Web Science and Protection of Privacy

health related information have come increased challenges to protection of individual rights to privacy.

6.2

Protection of Electronic Health Data on the Web

Information and communications technology may provide many benefits for patients in health care delivery: reducing chances of misdiagnosis, easier monitoring for chronic illnesses, streamlining wait times, providing ease of access to previous test results, and providing better care for patients in remote areas. Electronic health records are intended to be shared online across different healthcare organizations. The Web allows for instant access of a wide diversity of data and information, but in some cases people may wish to keep personal health information confidential. We discuss the current state of health privacy on the Web here. The language of privacy typically surrounds the actions that can be taken in respect of information: collection, use and disclosure. In respect of online personal health information, each is discussed in turn below, including a brief description of the current mechanisms used to address these requirements.

6.2.1

Health Data Collection

The collection of health information by organizations in any manner (including online) is generally guided by health privacy legislation that sets out two key requirements: authority and purpose. Authority requires that the collection of medical information be authorized in some manner, either by law or by some form of consent. Purpose dictates that the organization collecting the information sets out why they are collecting it, and sets the limit of the collection to that required in order to fulfill the purpose. On the web, limiting collection can pose a challenge. Electronic documents often have metadata that may or may not be easily visible to the patient or the organization collecting the data. Online submission methods, e.g., email or forms, carry that metadata through to any future recipient of the information. For example, if I submit an email to my physician, s/he is collecting not only the content of that email

6.3 Health Data Use and Users of Health Information

369

but also my IP address and message header information generated by the machines that support the collection (sending of the email) process. Health privacy legislation typically requires that organizations restrict collection practices in different ways: expressly identify the purpose for the collection of data at the point of collection; e.g., in a consent form or notice of collection, a patient is informed of the possible uses of their health information by the organization including fundraising and research; encouraging organizations to collect medical information directly from the patient, as opposed to from other healthcare organizations or family members (called secondary collection); inform patients in a consent form or notice of collection of the other organizations that may have access to some or all of their medical information; and prohibit any means of collection that is misleading. It may be quite difficult for a healthcare organization to provide a complete data flow to patients at the time of collection, for example, in an emergency room setting. Online applications also make it difficult for any covered entity to anticipate organizations that may end up with access to medical data through the entire information lifecycle, e.g., marketing companies working on behalf of pharmaceutical companies that access prescription records.

6.3

Health Data Use and Users of Health Information

Health privacy legislation in most jurisdictions outlines permitted uses, with the clause that other uses may be acceptable with express or implied (depending on the type of use) consent by the patient. When the patient information is held online two particular considerations arise: (a) the granularity of the consent and (b) the volume of data. EHealth software applications and databases can contain hundreds of thousands of data elements linked to individuals, including basic demographic data (referred to as tombstone) and all diagnostic information, e.g., blood type, HIV status, treatment options. Depending on the type of health services provided, a patient’s electronic record may also include information they entered themselves (such as a patient health record, or PHR) and/or information from community and social services provided, e.g., mental health status, eligibility for disability,

370

Health Web Science and Protection of Privacy

Fig. 6.1 Legal users of medical records under HIPAA.

discharge planning. Patient Privacy Rights (an American health privacy advocacy group) developed a diagram to demonstrate the number of organizations legally permitted to access this data (Figure 6.1). Under the Health Insurance Portability and Accountability Act (HIPAA), there are four categories of covered entities (Zone 2), including prescription drug card sponsors, health plans, clearinghouses, and certain health care providers. These entities may use personal health information without consent, and number in the thousands if not millions. “As a consequence, there is a huge market for health records in the US, and apparently 35% of Fortune 500 companies have admitted using purchased medical records for hiring and promotion purposes” [6]. Most health privacy legislation has initiated a reactive response by healthcare organizations. Primarily, policy and accountability structures are incorporated first, for example, by appointing a privacy officer. Research in this area typically focuses on business solutions. The

6.4 Disclosure

371

most common of these include: privacy policies; privacy impact assessments; logging and audit; privacy breach protocols, and incident management programs. These types of solutions target the end user, and apply within specific organizational frameworks. The nature of web often conflicts with the traditional organizational accountability structure set out in health privacy legislation. For example, a patient publishes their medical information on a social networking site. One of their health care providers engages in a discussion on that same site with that patient. Organizational boundaries do not exist in this environment. It is difficult to apply legislation that hinges on the ‘offline’ world for information exchange.

6.4

Disclosure

In privacy, the act of disclosure applies in two circumstances (a) when an individual makes information about them known, or (b) when an organization makes known information about an individual. Health privacy legislation applies to a variety of health care organizations and health information, and permits them to make disclosures of health information in a variety of situations, including back to individual themselves; for treatment, payment or health care planning; with consent for specific circumstances; for public interest (including research or public health); and when de-identified.7 The issue of consent is perhaps the trickiest when it comes to web sciences and health information. For example, if I post about a negative HIV test result on Twitter, am I making that information publicly available? In some legislative regimes, the answer is yes. In others, the answer is not so clear and often depends on the organization that is ‘collecting’ that information. Advocates of ‘publicness’ of medical information suggest that the benefits of sharing for a patient outweigh concerns, citing learning about correlations in diseases and developing support communities as advantages.

7 House

of Commons Health Committee. The electronic patient record. Sixth report of session 2006–7. Stationery Office, 2007.

372

Health Web Science and Protection of Privacy

6.4.1

Implementation and Enforcement of Health Privacy Legislation

Sovereignty and the rule of law are principles that allow physical jurisdictions to create and establish laws to codify behavior. In the US, as in most western countries, privacy is one such behavior. Regulated by a variety of laws that apply at different levels of government; municipal, state, and federal. Each law applies to different combinations of organizations, services, and data types. For signatories to the United Nations Declaration of Human Rights, rights to privacy are provided in Article 12, which states that no-one shall be subjected to arbitrary interference with his privacy, family, home, or correspondence, nor to attacks upon his honor and reputation, and that everyone has the right to the protection of the law against such interference or attacks. Table 6.1 provides an overview of health specific legislation. There is other privacy legislation that may apply to personal health information (defined in statute) or health information generally; this table is restricted to Acts that are specifically in respect of information management and privacy in healthcare. The Internet, however, is international; a place where traditional notions of sovereignty cannot apply. In this online ‘place,’ there are a number of direct and indirect privacy considerations, including mechanisms for collection, consent management, right of access, use (primary and secondary), data security and storage, general policy statements

Table 6.1. Country USA USA USA Canada Canada Canada Canada Canada Canada Canada Canada

Health privacy laws.

Region Federal Federal Federal Province Province Province Province Province Province Province Province

— — — — — — — —

British Columbia Alberta Saskatchewan Manitoba Ontario New Brunswick Newfoundland and Labrador Nova Scotia

Law HIPAA HITECH Financial Services Act PHIAPPA HIA HIPA PHIA PHIPA PHIPAA PHIA PHIA

6.4 Disclosure

373

(end user license agreements and acceptable use policies), training, and education. Nor can privacy be assured by governments acting alone. Governments, the public sector, and private actors each have roles to play in the protection of medical privacy. In the online environment, the protection of health data transcends the traditional approach of legislation in each sovereign jurisdiction. Already complex, the challenge is made more difficult by the need for multidisciplinary approaches that consider healthcare practices, information management, technical architecture, and requirements for compliance presented in different privacy regimes. Health privacy legislation is typically enforced by regulatory bodies, which publish various training, awareness, and educational materials to encourage individuals to consider the implications of sharing health information prior to disclosure. See, for example, the Office of the Privacy Commissioner, of Canada, Federal Trade Commission, and the Department of Health and Human Sciences in the United States. A number of advocacy groups also provide information intended to inform the public about the implication of sharing personal health information online, including Patient Privacy Rights and the newly formed Coalition for Patient Privacy. Older privacy advocacy groups have health privacy domains, including the Center for Democracy and Technology, Privacy Rights Clearinghouse, Electronic Frontier Foundation, and the Electronic Privacy Information Center. The next section examines the relationships among the different constellations of actors that influence policy and regulation surrounding health and the Web, and the consequences of these relationships for Health Web Science and Health Web Science researchers.

7 Health Web Science and Health Policy

Why is it necessary and important to discuss policy when thinking about Health Web Science? First, governments have a large impact on our health. In almost every country in the world, governments fund healthcare and claim the authority to protect public health. They regulate health providers, license health professionals, and inspect food imports and working environments, among many other tasks. They are by no means the sole actors in these areas. But what governments do, and what they plan to do — in other words, government policy — matters a great deal for the health of individuals and society as a whole. Second, governments have an important role to play (sometimes positive and sometimes negative) in determining how new technologies are created, diffused, and regulated. Government officials, legislators, regulators, and in many cases, judges, claim the right to act in the public interest, and aim to shape how societies interact with technology based on that authority. Government policy on issues such as funding for research and education, subsidies for innovation through grants or taxation policy, consumer privacy, or licensing the broadcast spectrum

374

7.1 The Effect of the Web on Key Elements of Health Policymaking

375

have an important impact on the development of the Web. In terms of Health Web Science, government policies can promote or inhibit the development of electronic patient records, incentivize the dissemination and reuse of scientific findings to facilitate population health, regulate the integration of mobile health applications with health systems, or license technologies for remote patient monitoring or physician training, among many other potential activities. Contemporary decisions about both health policy and technology policy are being made at a time of great economic uncertainty. Health care systems in many countries are in crisis. Aging populations, the increasing prevalence of long-term conditions, the increasing effect of humans on their environment, and a “new era of austerity” regarding public finances mean that there is growing pressure to change the way health care and public health services are delivered. Economies are also in crisis, with many economies in recession or struggling to grow. Many policymakers are looking to technology, including the Web, to alleviate these problems by saving money or encouraging economic growth. There is therefore a great need for Health Web Science scholars to both understand and inform these policy debates. Bearing these dynamics in mind, this section asks two key questions. What is the effect of the Web on key elements of health policymaking? And what is the effect of the Web on health policy? We will now consider each of these questions in turn.

7.1

The Effect of the Web on Key Elements of Health Policymaking

Given the immense potential scope of government influence in Health Web Science, it is perhaps helpful to think of the policy dimensions of this field in terms of three key elements commonly used by political scientists to increase their understanding of political environments: (i) ideas, (ii) interests, and (iii) institutions [98, 132, 194, 21]. Each of these elements plays a role in how policy on health and the Web is developed, decided, and implemented, although this role can vary substantially among political systems.

376 7.1.1

Health Web Science and Health Policy

Health Policy Ideas

The idea of the Web is a powerful in and of itself. Policymakers and think tanks often invoke the Web as a symbol of progress, innovation, and consensus. Members of political parties with strongly opposing views on most matters will frequently claim to support the extensive application of Web technologies to healthcare and public health policy in order to achieve some stated goal. The Web, along with similar technological developments, is often viewed as what might be termed a ‘magic bullet,’ a solution to intractable health policy problems such as rising healthcare costs, deficiencies in the quality of care, or inability to maintain adequate public health surveillance (for perspectives on this issue, see [109], Amitabh and Skinner 2011, [156]). But the evidence base for such assertions is often less clear. It can be difficult to calculate the true costs and benefits of adopting webbased systems, or to generalize from one setting to another, due to the number of intervening variables (for example, contrast Refs.[77, 13]). Exacerbating this dynamic is the fact that there exists considerable information asymmetry [197, 11, 3] between the public at large, those deciding policy (mostly elected representatives) and those implementing it, such as regulators, private companies, or non-profit organizations (who are more likely to be subject matter experts) (for health examples, see Refs.[191, 104, 87]). The Web undoubtedly does hold a great deal of potential to alleviate many difficult policy problems, but discussions of the details of how this can be accomplished are often divorced from existing or potential bodies of evidence for action. One vital role for Health Web Science is therefore providing an evidence base for policy decisions that affect health, and communicating this evidence effectively to decision makers. Attempts by governments to publicly disclose, or otherwise make more widely available, large health data sets could have a significant impact upon the evidence basis for future health policy. 7.1.2

Health Policy Interests

In the fields of public policy and political science, scholars like to talk about ‘interest groups,’ ‘pressure groups’ or ‘stakeholders,’ in other

7.1 The Effect of the Web on Key Elements of Health Policymaking

377

words, those organizations with an vested interest in a particular policy and how it plays out in practice ([209, 106], Baumgartner and Leech, 1998). Because healthcare and public health are life or death issues, and because governments (even in extensively privatized health markets such as the United States or Mexico) have such a large stake in health systems and services, the interest group politics of health are particularly contentious, meaning both that the fundamental issues are of high salience and that in every country (and internationally) there exist multiple, health-engaged interest groups which contest this policy space [93, 220, 200]. Many of these interest groups, such as health providers, pharmaceutical manufacturers, associations of medical professionals, and public health advocacy groups, derive profit from the health policy decisions made by governments, or feel that they represent a key segment of society, and thus engage in extensive lobbying efforts to alter policy outcomes. But individual members of the public are stakeholders also. One of the greatest potential strengths of the Web in health is to level the playing field somewhat between interest groups with considerable resources, those with few resources, and individuals. The Web can do this in two ways. It can provide health information to those who may not have previously been able to access it [62]. And it can provide a medium for individuals to form groups and establish relationships with one another in ways not previously possible [5, 140, 114]. Taken together, these two factors can change the balance of power considerably between groups, for example, empowering patients or public health activists. The formation of an international advocacy coalition to support access to essential medicines is a good example of the Web’s ability to support political mobilization on health issues. From the late 1990s, non-governmental organizations, activists, and medical professionals in countries around the world came together to push for better access to diagnostic tests and anti retroviral medicines to identify and manage HIV/AIDS, challenging both national and international policies in health, development, and trade (for an overview of the issue, see Refs.[219], M´edecins Sans Fronti`eres, 2009, [178]).

378

Health Web Science and Health Policy

Like many similar social movements, the campaign used the Web extensively to communicate with potential allies, to communicate with other members of the movement, and to frame the movement’s arguments in the global and national media. The Web does not cause such transnational social movements to form, nor were they non-existent before its formation. But in political campaigns like this one, it can potentially (to the extent that participants can access the Web in the first place) reduce the costs of participation, assist in promoting group identity, and facilitate relevant information sharing (for a full discussion, see Garrett, 2007). 7.1.3

Health Policy Institutions

Third, the institutional environment in which policies are formulated, decided upon, and implemented has a very important effect upon the development of the Web and its application to the field of health. In the social sciences, institutions are bodies of well-established rules and norms that govern collective behavior in society. They can be more or less formal, ranging from legislatures and agencies to sets of established guidelines [161, 99]. One key factor that alters how effective and appropriate policies are is the distribution of government authority among different institutions. This can be a source of tension in regulating the Web, as while the real world is extensively subdivided, information and relationships on the Web often do not respect these boundaries. Difficulties may arise in areas such as eHealth, where healthcare payers, service providers, patients, and patient data may exist in different political or legal jurisdictions. Such circumstances, where no one government body has authority to regulate or where the boundaries of that authority are untested or unclear, and where medical supply chains cross multiple national borders, may make defending the public interest more difficult. There are some examples of crossborder governance that demonstrate both this problem and the attempts to solve it. The European Union’s attempts to make policy on eHealth are one such example. The European Union (EU) is a common market and set of supranational

7.2 The Effect of the Web on Key Dilemmas in Health Policy

379

political institutions that pools the sovereignty of 27 countries. A key point of contention among those countries is frequently how to allocate decision-making authority between national governments and the EU’s central institutions. Under the EU’s founding treaties, health services remain a responsibility of national governments, who define health policy, and manage and finance health services (Treaty on the Functioning of the European Union (TFEU) 2011). In crossborder issues such as eHealth, the EU’s central institutions lack the authority to make policy on behalf of the EU as a whole, meaning that progress must be achieved via cooperation among national governments. Although most EU states have articulated policies on eHealth, implementation of these policies is patchy and depends greatly on national political will [126]. This dilemma is one reason why policies to support the creation and dissemination of eHealth applications are weak in crossborder contexts (Jarman 2013, Mars and Scott 2010).

7.2

The Effect of the Web on Key Dilemmas in Health Policy

With these dynamics of the policymaking process in mind, we can identify, and critically assess, three aspirations for the application of the Web to tough health policy dilemmas that are frequently stated by policymakers: hopes that the Web can reduce costs, hopes that the Web can produce improvements in the quality of healthcare and public health surveillance, and hopes that the Web will increase access to health information and thereby facilitate better health. The success of any of these policy ideas is highly contingent on the context in which the policy is formulated, decided upon, and implemented, particularly on the pattern of interests and institutions in each case.

7.2.1

Health Policy Costs

The first aspiration of policymakers advocating use of the Web to improve health is that health services could be made more efficient. Over the last 50 years, health spending (which includes spending from both public and private sources) in industrialized countries has

380

Health Web Science and Health Policy

tended to outpace growth in gross domestic product (GDP). Across the Organization for Economic Co-operation and Development (OECD) member countries, health spending was under 4% of GDP on average in 1960, but had risen to 9.6% by 2009 (OECD 2011, p.9). Increased public expenditure on health makes up a large proportion of growing health costs in the majority of countries, and attempts to curb health spending are highly contentious. It is possible to reduce government spending on health services by rationing care, closing facilities, or cutting salaries or benefits for the medical workforce, to give a few examples. But these solutions, which involve the substantial reallocation of resources, can be both highly visible and vastly unpopular, causing significant problems for politicians who must win votes at the next election. The development of the Web provides policymakers seeking to cut health spending with an additional set of ‘solutions’ to the health spending problem that may be viewed as more rational, scientific, or progressive than explicit resource reallocation, and therefore less contentious. The extent to which web technologies can and do reduce healthcare costs is difficult to judge. On the one hand, there is increasing evidence that where web-based systems can help to emphasize preventative care, they can result in cost savings. On the other, there is some emerging evidence that electronic health records might increase costs, rather than reducing them. A lack of trust in online systems among physicians may cause them to re-order tests instead of basing their decisions on existing patient information, for example. Electronic billing and record keeping may reduce administrative costs for health providers, but there is no guarantee that providers will pass these cost savings on to healthcare payers. In some cases, electronic records may actually enable health providers to charge more by streamlining the payment process and making it more transparent. Despite this mixed picture, governments have gone to great lengths to incentivize use of the Web in healthcare. Web applications may indeed hold the potential to reduce health spending through increased efficiency, but they are parts of complex health systems and thus cannot be viewed as stand-alone solutions.

7.2 The Effect of the Web on Key Dilemmas in Health Policy

7.2.2

381

Health Policy — Quality of Health Care and Outcomes

A second aspiration is that the dissemination and exchange of health data, facilitated by the Web, can improve health outcomes by improving the quality of healthcare or of public health surveillance. Many industrialized states have made substantial improvements in the quality of their healthcare in the last two decades, although differences between states persist. For example, the US, UK, France, and Germany all reduced their rates of amenable mortality (deaths which could potentially have been prevented if patients had received timely access to appropriate care) between 1999 and 2007. But while the rate declined by 37% in the UK, 28% in France, and 24% in Germany, the decline in the rate was only 18.5% in the US [159]. There are several potential benefits in health quality that are frequently attributed to the use of Web technology, including better monitoring of patients (after they are discharged, or to track ‘hidden health behaviors), or improved aggregation of information in ways that could promote patient safety, such as reducing medical errors, providing test results more rapidly, providing drug alerts, or increasing continuity of care [165]. There are also indications that the Web could facilitate diagnosis and treatment that is based on the latest evidence. The sheer volume of information on any given health topic can be overwhelming. Even with the help of systems to facilitate filtering and classifying this information, it can be difficult for medical professionals to keep up with the latest medical evidence. What policies exist to promote the use of web technologies to improve healthcare quality? In the past few years, governments in several high-income industrialized countries have started to promote the collection and use of ‘big data,’ to promote more accessible benchmarking against targets, or promoting government data disclosure and reuse. The US ‘data.gov’ experiment, a government sponsored open-access repository and ‘data community,’ is a good example of this trend. One difficulty encountered by data.gov, and its health-related counterpart, healthdata.gov, is how to design policies that promote

382

Health Web Science and Health Policy

the use and reuse of existing data or other research findings that could impact healthcare quality. As researchers have discovered, making more information available does not guarantee that it will be used. Not only that, but those using big data to make health policy decisions about funding or the effectiveness of interventions should be cautious: data is difficult to judge without context, and there exists a danger that policymakers will make decisions without understanding the implications or limitations of the data they are using. Government officials in the age of big data may well require a different skillset than previous generations of civil servants. 7.2.3

Health Policy — Access to Information and Services

A third aspiration is that the Web can improve access to health information and health services. Access to healthcare is fundamentally a problem of resources, whether this refers to the ability or willingness of governments to cover the costs of care, restrictions on the availability of medicines, the supply of medical professionals, the distribution of medical facilities, or the socio-economic status of patients. Among high income, industrialized countries, the United States has by far the worst healthcare coverage and affordability, although access to healthcare is a universal problem. A recent survey of primary care physicians in 10 high income countries found that 59% of the US physicians surveyed said their patients ‘often have trouble paying for care’ — compared to 4% in Norway, 13% in the United Kingdom, 16% in Switzerland, 21% in Germany, and 25% in Australia. Fifty-two per cent of US doctors surveyed said they or their staff spent ‘too much time dealing with insurers’ restrictions on covered treatments or medications’ [183]. It is not an accident that the United States is the only country in the survey without a health policy that promotes some form of universal coverage. The existence of the Web cannot erase resource disparities or income inequalities. But web technologies do have the potential both to reduce some of the barriers that these disparities create, and to empower individuals and communities to demand a better allocation of resources. For example, the WHO Observatory on eHealth believes that telemedicine

7.3 Informing Public Debates About Health and the Web

383

has the potential to overcome the geographical barriers to accessing healthcare that exist in the rural or otherwise underserved areas of developing countries [213]. There are many other ways in which the Web can potentially improve access to healthcare and public health services, including by providing information about how to get care, diagnosis and treatment, and guidance for self-care, by promoting ownership of treatment or preventative behaviors, or by promoting the formation of support groups and community identity.

7.3

Informing Public Debates About Health and the Web

The impact of the Web on our health is mediated to some extent by policymaking processes and vice versa. The effects of the Web on health policymaking include (i) catalyzing the dissemination of health policy ideas and evidence, (ii) altering the constellation of groups with vested interests in health, and (iii) by highlighting the limitations of national political institutions and legal jurisdictions in dealing with global problems. In this regard, governments and policymakers in health are discovering what many who study the Web already know: that the causal relationships between technology policy, health innovation, and economic development are far from clear. Historically, many attempts by states to foster the development of new technologies, the dissemination of new applications, or the commercialization of innovation, have been fraught with difficulty, and the Web is proving to be no exception. The effect of the Web on health policy is to provide an additional means by which persistent health policy problems, particularly rising healthcare costs, deficiencies in healthcare quality, and access to healthcare and public health services, can be addressed. But as a policy solution, the Web, like all policy solutions, is flawed. There are also many real world divides that the Web can sometimes replicate rather than alleviate. Not everyone can access the Web. Not everyone can access healthcare or preventative services. And even those with access may not have the resources to capitalize on the health information that the Web can deliver, either in terms of skills, knowledge, money, or time. Nevertheless, the Web, like radio, and television before it, has changed health politics and policymaking in three very important

384

Health Web Science and Health Policy

ways: government transparency, evidence formation, and group mobilization. As a result, many areas of policymaking are now more transparent. The Web facilitates scrutiny of government actions to promote health. Second, governments are releasing significant amounts of the health data that they hold into the public sphere, and encouraging researchers to share their data also. This collaboration could provide a much improved evidence base for future health policy. And third, individuals, groups, and communities can access more information about their own health and healthcare, as well as the health policymaking. The Web has shown great potential as a tool to support the mobilization of activists and advocates, the formation of pressure groups, and the construction of social movements. In many ways, the policy task for those promoting Health Web Science is to connect these three trends — government transparency, evidence formation, and group mobilization — in ways that promote better health for individuals and populations. 7.3.1

An Example — Scotland’s Strategy

The Scottish Government issued an eHealth Strategy in 2011 to cover the six-year period from 2011 to 2017. This followed on from and builds upon the eHealth Strategy that ran from 2008 to 2011. Nicola Sturgeon, the Deputy First Minister and Cabinet Secretary for Health, Wellbeing and Cities Strategy is quoted: “The potential of information technology to support and transform healthcare services is universally recognized. Here in Scotland, eHealth has a pivotal role in enabling a radical e-transformation in the way in which high quality integrated healthcare services are delivered efficiently and effectively to people of all ages across our country. Our vision for 2017, and this second eHealth Strategy for NHS Scotland, is ambitious; it has the citizen at the center and seeks to build on the significant progress we have made over the course of the last three years. Rather than focusing on products and technology, we will instead look to the benefits and outcomes experienced by the people of Scotland flowing from eHealth enabled service re-design and quality improvements.” The strategy identifies five aims and is framed around three quality ambitions: the strategy highlights

7.3 Informing Public Debates About Health and the Web

385

that people with long-term conditions are the major users of health and social care support services. Over two million people in Scotland live with one or more long-term condition and their occurrence is set to rise significantly with an aging population. Long-term conditions currently account for around 80% of general practioner consultations, which is unsustainable. eHealth can support people with better information and minimize unnecessary face-to-face contact with health services, essential if we are to meet the future challenges of an ageing population and reductions in health spending. The terms Telecare and TeleHealth are seen in the strategy elements of eHealth and as technologies can provide effective self management support tools for people with long-term conditions, which enable them to live independently at home with a much greater understanding of their condition and an improved quality of life. Giving people better information about their conditions has been shown to empower individuals (and their caregivers) and support self management. TeleHealth and telecare can be used to support the whole spectrum of long-term conditions and to develop, in partnership with health and social care workers and patients, potential areas for better self-management. The strategy asserts that Telecare is an important eHealth tool that supports people, particularly the elderly, or those with long-term health conditions, to live at home for as long as they want to. Up to 2011 telecare has been supported with £20 million from the Telecare Development Programme. Approximately 29,000 people have received a telecare service since 2006, and over 2000 of these people were diagnosed with dementia [211]. Scotland is currently performing well in the development and implementation of telecare and teleHealth services [185]. It is estimated that efficiencies (in the form of care home and hospital bed days saved/avoided admissions) of around £48 million have been made (Technology Strategy Board, 2011). 7.3.2

Policy on Quality Assurance of Health Information

The quality of health information available on the Internet is extremely variable and debate surrounds how to measure and ensure quality on the Web. In the UK, the British Standards Institution (BSI) group has begun to issue a Child Safety Online Kitemark (BSI logo) that

386

Health Web Science and Health Policy

indicates the parental control software has been certified to meet the highest standards thus assuring parents that their children can surf safely. Conversely, despite the availability of miscarriage-related websites on the Internet, a Scottish survey found that they were of poor quality, did not meet with the clinical guidelines produced by the Royal College of Obstetricians and Gynaecologists (RCOG), and had not been adequately assessed for therapeutic benefit in accordance with the requirements of evidence-based practice [101]. The Pew research study showed that 86% of health surfers are concerned about getting health information from an unreliable source on line, though 52% of users who had visited health sites thought that almost all or most health information on the Internet is credible [73]. Many researchers, organizations, and website developers are exploring ways of helping people find high quality information on the Internet [173]. Since 1996, many organizations have developed tools for rating health websites and Gagliardi and Jadad [79] concluded that many incompletely developed rating instruments continue to appear on health websites even though the organizations that gave rise to those instruments no longer exist. Wilson [215] described tools for rating the quality of health information on the Internet that include self-applied for quality labels, e.g., Health on the Net (http://www.hon.ch/HONcode/HON CCE en.htm#1) and externally applied for quality labels (e.g., star rating systems by users), gateways or filters (credible organizations that evaluate and then recommend health websites), and inbound links that endorse a website by linkage from another site.

8 Conclusion

“To him who devotes his life to science, nothing can give more happiness than increasing the number of discoveries, but his cup of joy is full when the results of his studies immediately find practical applications.” — Louis Pasteur In this monograph, we present Health Web Science (HWS) as a subdiscipline of Web Science that, while being interested in the Web’s impact on health and wellbeing, also examines the impact of the Web’s health-related uses on the design, structure, and evolution of the Web itself. Understanding and appreciating the overlapping, yet divergent disciplinary orientation of HWS compared to related research domains, motivates specific research efforts around better utilization of, innovation on, and communication over and within the Web. Health Web Science emerged as a result of the need to understand the interplay between health and wellbeing and the Web, through the academic lens of Web Science. The Web has affected, and been affected by, all aspects of health research and delivery and that there is a clear need to unite interests 387

388

Conclusion

germane to health-related uses of the Web under the rubric of HWS. HWS is, therefore, properly defined as a domain-specific subgroup within Web Science that seeks to understand and describe how the Web shapes, and is shaped by, medicine and healthcare ecosystems. Through this information, HWS will help engineer the Web and web-related technologies to facilitate health-related endeavors, in the broadest sense, and empower health professionals, patients, health researchers, and lay communities. Features relevant to HWS include, for example, the synthesis, curation, and discovery of web pages containing health information; the structure and utilization of interactive social media sites relevant to patient support groups; and semantic annotation and linking of health records and data to facilitate mechanized exploration and analysis. The focus of HWS is, therefore, more strongly allied with non-medical stakeholders compared with the interests of Health 2.0/ Medicine 2.0, whose scope reflects more the agendas of professional and particularly medical stakeholder [84]. The motivations to establish Health Web Science as a discipline are multi-fold. Increasingly, scientific breakthroughs are powered by advanced computing capabilities using the Web as a conduit; therefore, we must seek to understand and describe the distinct manner in which the Web is used for health research, clinical research, and clinical practice. In addition, we must seek to support consumers who utilize the Web for gathering information about health and wellbeing. We must elucidate approaches to providing social support to both patients and caregivers. Finally, there is the motivation to improve both the effectiveness and efficiency of healthcare where quality improvement and cost containment have become international priorities as governments, employers, and consumers struggle to keep up with rising healthcare costs. Health Web Science proposes that these motivations will be advanced through a comprehensive understanding of the current boundaries and properties of the health-related web, as well as designing novel ways to utilize and engineer the Web to maximize its utility as a health resource. Health Web Science is therefore integral to solving the health problems of the twenty-first century in both the developing and developed world. There is no perfect health model, however there will

389 be a shift from the current model of a centralized, hospital-focused and provider centric infrastructure to one where the hospital plays more of a co-ordination role and interacts with a long tail, distributed peer-to-peer model as wellness and self-care become the predominant paradigms driving health care delivery. Health care will become tailored to individual needs based not only on medical conditions but also on personal, family & social grounds with such relationships varying with time, and such variations being driven by metadata and studied by Health Web Science. Health Web Science will help to provide the evidence base of what works best where and when, and with what conditions, and for whom. The discipline will also leverage knowledge gained through its multi-disciplinary study to provide an evidence base for other disciplines to provide targeted interventions. Health Web Science’s unique niche lies in its raison d’ˆetre of describing and understanding the way people engage with, access and use the internet for health, and in particular, the Web and other web like structures vis-`a-vis, the web of linked documents, the social web, the web of linked data and apps. Furthermore health Web Science investigates and describes how that interaction not only changes the topology of the Web but how our interaction with the web for health changes us, as in the case of “the medium is the message” [142] or the “filter bubble” where what we see is pre-determined by a social-media algorithm [166]. These interactions can be unpredictable with unintended consequences (both healthy and unhealthy) and it is the remit of Health Web Science to describe and make known these emergent characteristics to policy makers, health professionals, and the public. In addition, it is being increasingly recognized that good health is also connected to our social networks. It is therefore not solely a matter of individual responsibility. Health Web Science needs to understand the ways that social networks can influence our health outcomes, for example, is obesity contagious via social networks? Some social networks are more prone to cascades or have nodes that are more effective at influencing the whole. This understanding therefore allows targeting in order to disseminate messages in the most influential manner. One of Health Web Science’s differentiation from the disciplines under the Health 2.0/Medicine 2.0 umbrella is scale with an emphasis on “big data.” This data exists in information

390

Conclusion

silos (although this is beginning to change with cloud technology) that need to be linked and interrogated. Health Web Science needs to show leadership at local, national, and international levels in driving the necessity for data liquidity as well as leading and informing the debate on overcoming the quality, privacy, provenance, security and ethical problems as they arise. The quantified self (self knowledge through numbers) whereby a person collects his/her own personal data from sensors/smartphones, and uses this information as a feedback loop for self improvement is gathering momentum and resonates with the wellness agenda and the need to empower people to look after and take responsibility for their health. Some of these apps then connect to social networks for additional motivational support, e.g., (RunKeeper). This quantified self is a natural progression from the current practice of monitoring patients either acutely or as part of managing a chronic illness (the medical and socially integrated quantified self). Smart/intelligent cities (using ICT to create networked infrastructure across the economy, mobility, environment, people, living, and governance) and Smart Villages (where houses are technologically enabled to assist independent living) will increase the amount of data available for analysis and interpretation. As organizations, social networks and data become increasingly connected utilizing the Web, Health Web Science has a role in the wellness agenda and enabling patients to remain at home and avoid the expense of hospital based care. Social web sites are encouraging “participatory cultures” whereby the public enter their own data on medical conditions (http://curetogether.com) and then using the economies of scale, use computational power to ask open questions and look for patterns that may offer new therapies or change practice. This culture is based on the fact that the Web makes it easy to find those with similar conditions and then be able to share insights and together search for a cure (citizen science) and also act as a pressure group to demand better products and services to manage their condition. Citizen science is the reverse of the current scientific epistemology of evidence-based medicine (EBM) that is regarded as glacial in speed of obtaining results in comparison. The use of metadata rather than recruitment of subjects for clinical trials could streamline the process for EBM and improve the speed of obtaining results. This form of

391 population mining/fishing/evidence gathering has also been regarded as full of noise and poisonous by the EBM community (but, is this any different from the resistance and now widespread use of microarrays in molecular biology?). In contrast, EBM is the conscientious, judicious, and explicit use of current best evidence in making decisions about the care of individual patients (Sackett BMJ 1996). Nevertheless web speed research has shown promise. Looking for an association between Gaucher’s disease and Parkinson’s Disease using traditional research methodology took 6 years whereas using the Web took 8 months with similar results. PatientsLikeMe are also sharing and distributing medical data and sharing their own research and publishing in journals challenging lithium’s efficacy in the treatment of Amyotrophic Lateral Sclerosis. If individual searches or websites can achieve this success, the use of metadata might promote novel treatments for disease, or even define whole new disease processes. Health Web Science needs to understand the strengths and weaknesses of both types of research methodologies and when to apply them. Health Web Science needs to take a lead in solving the problems of interoperability and designing the architecture of the internet and the web to enable connected health. Tools need to be developed and made readily available to not only capture and curate this “big data” but then to turn this data into wisdom which is accessible to the public (data being raw material from which meaning is derived, information being data that’s been collected and organized, knowledge being information we have digested and understood and wisdom being the proper use of knowledge). The availability of this wisdom will become increasing instantaneous (the healthcare singularity), i.e., whereby information from bench to bedside is immediate and is understood. This is a debatable point and it is the reason why health professionals undergo a long period of trainingit’s what Aristotle called “phronesis” or practical wisdom. It is because the logic of care is unbounded, non-linear and unpredictable that there cannot be an algorithm for every situation. It is why humans reason differently from computers. Health Web Science needs to be involved in this debate as proponents of artificial intelligence argue that these problems can be overcome. Health Web Science only becomes credible if it can deliver the above aims. In doing so it must cross the economic,

392

Conclusion

social, and digital divides and successfully influence political agendas at policy level to counter the claim that many feel that information and communication technologies have overpromised and under delivered. In Section 1 we described how the impact of the invention of the Internet and web has potentially greater impact on shaping society than Gutenberg’s printing press. The widespread use of the printing press was game changing, it was disruptive, and it was the catalyst that led to the stepwise democratization of power from the elite to the many — a situation we now recognize in the term — the wisdom of the crowd. It put the individual at the center and it is no exaggeration to say Copernican in its impact by opening up the possibility of challenging these top-down hierarchies. The genie was out of the bottle. However, it was not until the widespread adoption of the personal computer and easy access to the internet/web that this revolution that started in Germany really gained traction. In the last 20 years, an individual’s potential to influence society has increased exponentially. The democratization of health has seen its outworking in the changing doctor/patient relationship (from a doctor knows best paternalistic model through an informed model to a shared model). This has been achieved through the Web of linked documents and the social web has seen the growth of patient power and the increased accessibility of patient groups with health professionals. Experiences both positive and negative can be shared and these can be correct or incorrect. Personalized medicine is going to come from the patient, not the practitioner. “Expert patients” are taking control of their own health information, educating themselves, networking with one-another, and even driving the direction of spending the public purse on clinical trials in Canada to pay for venoplasty as a treatment for multiple sclerosis. A treatment no clinician wants to support, as the problem is autoimmune in origin. Expert patients will soon be able to sequence themselves, thereby taking control of their genomic information out of the hands of their care-providers. Unfortunately, the Internet is a vast trove of both valid medical information, as well as utter fiction; moreover, the information on the Web is, by-nature, generalized, and even valid information may or may not be relevant or applicable to any given patient. Health Web Science must therefore, find novel ways

393 of providing relevant, accurate, personalized, expert medical information, and evidence-based guidance to patients as they manage their own health-care. To do otherwise may result in great harm. Not only are patients dealing with information overload, clinicians are also struggling to keep up-to-date with the medical literature and best practice despite specialization and sub-specialization. For instance, it is estimated that a physician in epidemiology, needs 21 hours of study per day just to stay current. One hundred years ago, a physician might have reasonably expected to know everything in the field of medicine. Today, a typical primary care doctor must stay abreast of approximately 10,000 diseases and syndromes, 3000 medications, and 1100 laboratory tests; the list grows every year [83]. To mitigate the problem of information overload, Semantic Web service protocols and tools are required to allow workflows to be automatically synthesized, based on a combination of text mining for semantic concepts within a page, ontologies, and relevant databases. The analytical workflow that provides the page mark-up will potentially be different for any given user because it will be based on the databases and the ontologies being used to interpret that data. The speed of change is exponential; an exemplar being Moore’s Law that states that the number of transistors on integrated circuits doubles approximately every two years. Kurzweil [125] uses the phrase “the law of accelerating returns,” to describe this and similar phenomena, an argues that if a technology is successful then more resources are deployed toward the further progress of that process. Thus, a second level of exponential growth results; i.e., the rate of exponential growth, the exponent itself, grows exponentially. This speed of change has led to talk about a “healthcare singularity.” The word singularity comes from the field of astrophysics and refers to a point in space–time where the rules of physics do not apply. The healthcare singularity refers to a tipping point, where the flow of medical knowledge from research to practice (bench to bedside; innovation to implementation) becomes instantaneous [83], and because “the rules may no longer apply,” the future healthcare landscape beyond this point of is speculative. In practice, one can imagine that for some medical knowledge, such as a fatal side effect, immediate propagation

394

Conclusion

across the Web is both desired and appropriate, while other medical knowledge, for example, knowledge relating to long-term safety effects, would and should remain conservative and as such, take an appropriate time before it is put into common medical practice. Hippocrates (circa 460 BC–370 BC) is generally regarded as the founder of Western Medicine. In this monograph, we notice we have come upon a juxtaposition of the aphorism credited to him “first do no harm” with a modern equivalent (that particularly resonates with the nascent discipline of Health Web Science), namely Google’s mantra “do no evil”. David Wootton [218] argues in his book Bad Medicine that “the Hippocratic” tradition of medicine has caused more harm than good until the last 150 years. He offers us a history not of progress but of delay. This delay has been a result of a number of reasons some nefarious, others legitimate, of willful blindness where interests of big business or reputation and the conservatism of the medical hierarchy have delayed, obfuscated or at worse sought to discredit the evidence for a change in medical practice. It has also been because communication of the idea has been difficult due to limitations of the communicating tools. Health web science is positioned to help mitigate the bottleneck in the translation from bench to mainstream clinical practice currently taking 15–20 years. Morris et al. [150] argue this to be a crude approximation that hides many complexities. Nevertheless, the trend of this lag time from discovery to adoption by the medical community is decreasing, appears to be linear, and is projected to be “instantaneous” at 2025 (see Figure 8.1, [83]). It appears we are where we should be, on target, as predicted. Using ICT, Health Web Science therefore, has an integral role to play in explicating the web aspects contributing to a personalized, predictive, preventative, and participatory medicine. These contributions occur in the context of the technological intersection between medical experts, expert patients, and increasingly rapid knowledge dissemination. It has the potential to unlock the secrets of big data within a carefully controlled governance framework and help in separating web fact from web fiction and can transform the generalized nature of web information making it relevant and applicable to an individual patient. HWS is the pursuit of novel ways to provide relevant, accurate, personalized, expert

395

Fig. 8.1 “While it took 2300 years after the first report of angina for the condition to be commonly taught in medical curricula. . . The trend still appears to be linear, approaching the axis around 2025.” From: “The Healthcare Singularity and the Age of Semantic Medicine,” The 4th Paradigm. (Used with permission).

medical information, and evidence-based guidance to patients as they manage their own health-care. The momentum behind this emergent discipline has been established, and dialogue now needs to continue so that the various stakeholder communities can mutually educate each other for the greater benefit of society.

396

Conclusion

Explanation of Terms eHealth eHealth is an emerging field in the intersection of medical informatics, public health, and business, referring to health services and information delivered or enhanced through the Internet and related technologies. In a broader sense, the term characterizes not only a technical development, but also a state-of-mind, a way of thinking, an attitude, and a commitment for networked, global thinking, to improve health care locally, regionally, and worldwide by using information and communication technology. Telecare The term ‘Telecare’ refers to a range of equipment and associated services that support and enhance safety for people living at home. Telecare equipment mainly consists of basic community alarms with pendants but also includes devices that automatically trigger a response from a third party where a risk to the service user is detected. Telecare can also be used to prompt actions from service users, for example, to take medication and can be used for the capture of information

397 related to behavioral patterns as part of assessment and monitoring processes. Telecare equipment can be programmed to alert either a live-in or close-by caregiver or a 24-h monitoring center whose trained operators then determine the best course of action following dialogue with the service user. Telehealth Telehealth is the remote electronic exchange of personal health data from a patient at home to healthcare staff at hospitals or similar sites to assist in diagnosis and monitoring. The main benefits are that long-term conditions such as high blood pressure, diabetes and chronic obstructive pulmonary disease can be monitored remotely without the need for the patient to attend traditional healthcare settings. The patient is taught to take their own regular health readings and feed them back to the surgery using existing ‘phone lines or wireless technology.’ Telehealth can include the use of general Information and Communications Technology to promote better management of long-term conditions through, for example, prompting actions for self-care at times of risk of exacerbation. Telemedicine Telemedicine refers to real-time consultation between a patient and a clinical adviser by way of video link, web cam, or something similar. The benefits are that clinical decisions can be made quickly in situations where it would not be possible to get a patient to the clinician, for example, because of distance. Telehealthcare Telehealthcare can be used interchangeably with the terms Telecare and Telehealth. The term, which has been adopted by the Scottish Government, relates to the convergence of the two to provide a technological and integrated approach to the delivery high quality health and social care services. This convergence is becoming more apparent in Scotland generally with the expansion of, for example, monitoring of long-term conditions as part of a telehealthcare package in a person’s home. The following diagram shows key opportunities for convergence.

398

Conclusion

Teleheathcare — areas of Telehealth & Telecare convergence from 2010-2015.

Acknowledgments

This work was partially supported by the National Health Service Grampian, the University of the Highlands and Islands, the Center for Plant Biotechnology and Genomics at UPM/INIA, Case Western Reserve University, Predictive Medicine, Inc., and the Tetherless World Constellation at Rensselaer Polytechnic Institute. The ideas presented at Medicine 2.0 and in this manuscript have benefited greatly from contributions of content and discussions with the following people: Christopher J. O. Baker, Doug McKendrick, and Marie Joan Kristine Gloria. We are particularly thankful to David Molik for his editorial contributions and for acting as project manager to move us along at a critical stage. And lastly, thanks to Richard Seltzer, an original Internet Evangelist, who eagerly jumped lat the end to rescue the beginning text.

399

Appendix A Brief History of the Web

Today, three quarters of adults in the US use the Web, and 80% of them use it as a resource for health-related information Those health-interested information seekers in the US tend to be white (non-Hispanic), young-adult women with relatively high education and income1 Other industrialized countries also have heavy Internet usage and high interest in health information, but not as great as in the US.2 In this section we provide a brief history of the Web.3 The invention of Gutenberg’s printing press in the 15th century was responsible for providing impetus to the Protestant Reformation, the Italian Renaissance, and the birth of modern science. The political, 1 Fox,

Susannah, Health Topics, February 1, 2011, Pew Internet Project and the California HealthCare Foundation. http://www.pewinternet.org/∼/media//Files/ Reports/2011/PIP Health Topics.pdf 2 The BUPA Health Pulse 2010 International Healthcare Survey found in the 12 countries surveyed (Australia, Brazil, China, France, Germany, India, Italy Mexico, Russia, Spain and the United Kingdom) about 60 percent of Internet users searched for health related information online. https://www.bupa.com.au/staticfiles/Bupa/HealthAndWellness/ MediaFiles/PDF/LSE Report Online Health.pdf 3 The precurser of the Internet was the ARPANET (1967), a U.S. government network (Advanced Research Projects Agency Network). ARPANET was the network that became the basis for the Internet.

400

401

Fig. A.1 The impact of the personal computer and Internet on an individual’s potential to influence society. Depicted is the number of people (x-axis) that one can reach with the written word as a function of time (and communication technology), controlling for user acceptance, and accounting for disruptive technology.

scientific, and arts landscape of Europe as a result was irreversibly changed. The Internet is the modern day equivalent of the printing press and commentators are arguing that its effect on society will be equally as revolutionary [56], not least in an individual’s potential to influence society (Figure A.1). Health Web Science identifies the gamechanging impact of the Internet made viable through the Web and its potential to change health behaviors and health care worldwide. In health, this revolution’s impact on changing the practice of medicine can be considered in three areas: power, experience and speed (Figure A.2) and these themes will be expanded throughout the monograph. Claude Shannon’s seminal paper “a mathematical theory of communication” [186] marked the birth of information theory (he gave us the “bit” — the basic unit of information) that underlies this digital revolution [96]. Now, just over 60 years later, Pulitzer Prize nominee Nicholas Carr framed this sentiment for our age by asserting that “with the exception of alphabets and number systems, the Internet may well

402

A Brief History of the Web

Fig. A.2 The Internet revolutions impact on changing the practice of medicine.

be the single most powerful mind-altering technology that has ever come into general use [40]. Carr further argues that the way our minds interact with the Internet can have a profound effect on the neuroplasticity of the brain which could result in its “rewiring”. These ideas resonate with the ideas of Marshal McLuhan that “the medium is the message” and that “we shape our tools and then our tools shape us” (McLuhan 1964). McLuhan hypothesized that a medium affects the society in which it plays a role not only by the content delivered over the medium, but, analogously, by the characteristics of the medium itself. This change may be societal and individual. Carr rephrases McLuhan’s dictum “as not only do we create the Web but the Web has the potential capability to change us (re-create us).” During this same time, communication scientists in the US and the UK were developing the idea of “packet switching.” Packet

403 switching is a method to deliver data across a computer network that involves packaging data, regardless of content, type, or structure, into specially formatted units called packets. The packets, along with the communication protocols, i.e., the rules for exchanging data between computers, enables the packaged data to be routed on different paths and reconstructed at their destination. The TCP/IP protocols developed in the 1970s made it possible to expand the size of a network to what it is today.4 The set of standards and technologies that have come to be known as the World Wide Web were initially deployed in December of 1990 [27], and became publicly accessible in early 1992. A decade later, the Commission of the European Communities established a set of criteria for Health Related Websites (2002) in order guide users towards useful and reliable online medical and health information.5 The Web’s penetration became greatly accelerated with the advent of the Web Browser (1993) and the development of search engines (1997) [210]. Together these made information on the Web accessible to people outside the computing industry, but this was almost exclusively for consumption — the owner of the Website created, and provided, the content. This uni-directional flow of information began to change in the early 2000’s with the emergence of WikiPedia, MySpace, and Facebook. With these (and other) “Web 2.0” sites, individual users began to create the content of the Web. This, in turn, led to the ‘recommendation age’ [8], which saw the development of software that observed, recorded, and aggregated the behaviors of Web users in order to improve the experiences of other Web users. In typical recommender systems [172], people provide recommendations as inputs, which the system then aggregates and directs to appropriate recipients: for example, the ‘customers who bought this also bought’ service of online retailer Amazon. All of this led to Time Magazine naming the individual, i.e. “You”, as the 2006 person of the year, reasoning that the layperson was now able to control content on the web [95]. This trend only intensified with the 4A

detailed history of the developments leading up to today’s Internet can be found in the book Where Wizards Stay Up Late, by Katie Hafner and Matthew Lyon. 5 The major sponsors of Health On the Net Foundation are the State of Geneva, Geneva University Hospital, the Swiss Institute of Bioinformatics, and Sun Microsystems.

404

A Brief History of the Web

expansion of social media. These relative newcomers to social communication are now highly influential in shaping social interactions through the Internet, with Facebook and Twitter dominating social networking [189, 100]. In parallel with these technologies becoming mainstream, coproduction, and co-creation have become important philosophies underpinning business development strategies. These essentially synonymous terms refer to active input by the people who use services, as well as those who have traditionally provided them [20, 61, 167, 221]. It is here where we can begin to discuss the topic of this book since, as a result of the emergence of this constellation of technologies and behaviors, the Web has begun to intersect with the lives of both health professionals as well as patients, and at that intersection, Health Web Science is born. Putting patients at the center of healthcare, much as individuals are at the center of Facebook; embracing new ways of working where health services are more responsive to patients and designed around them, rather than patients having to fit around services (DoH 2010). Postman [168] argued that technological change is transformative; “In the year 1500, fifty years after the printing press was invented, we did not have old Europe plus the printing press. We had a different Europe. After television, the US was not America plus television; television gave a new coloration to every political campaign, to every home, to every school, to every church, to every industry.” The Web, without question, is having a similarly transformative impact. In twenty years, we have moved from the Web as a collection of linked documents using read-only technologies, sometimes called Web 1.0, to the Web as a medium of information exchange, sometimes called the Social Web or Web 2.0, which incorporates both read and write technologies [163].6 6 It

is important to note that the boundaries between these terms are not always clear. The concept of web generations was pioneered by Tim O’Reilley, but the term was coined by Darcy DiNucci. The term Web 1.0 was coined retroactively during the onset of the social web. The term Web 3.0 does not have a single author but was most likely employed by Semantic Web. The semantic web (Berners-Lee, Tim; James Hendler; Ora Lassila (May 17, 2001). “The Semantic Web”. Scientific American Magazine. Retrieved March 26, 2008.) came about before the concepts of linked data as marketing for the technologies they were trying to develop (Tim Berners-Lee (2006-07-27). “Linked Data—Design Issues”. W3C. Retrieved 2010-12-18).

References

[1] L. C. Abroms, N. Padmanabhan, L. Thaweethai, and T. Phillips, “iPhone Apps for smoking cessation: A content analysis,” American Journal of Preventive Medicine, vol. 40, no. 3, pp. 279–285, March 2011. [2] S. A. Adams, “Blog-based applications and health information: Two case studies that illustrate important questions for Consumer Health Informatics,” International Journal of Medical Informatics, vol. 79, no. 6, pp. e89–e96, 2010. [3] G. A. Akerlof, “The Market for ‘Lemons’: Quality Uncertainty and the Market Mechanism,” Quarterly Journal of Economics, vol. 84, no. 3, pp. 488–500, 1970. [4] H. Akkerman, N. B. Gyan, A. Bon, W. Tuyp, A. Grewal, S. Boyera, M. Allen, and S. Eco, “Is (Web) Science Ready for Empowerment?,” http:// www.w4ra.org/sites/default/files/documents/W4RA article-ACM WebSci 11 Koblenz-Jun2011-HansA-final.pdf, 2011. [5] C. J. Alexander and L. A. Pal, Digital Democracy: Policy and Politics in the Wired World. Toronto: Oxford University Press, 1998. [6] M. Altheim, “Computers, Privacy and Data Protection International Conference: European Data Protection,” In Good Health? Part 1. January 30th, 2011, http://ediscoverymap.com/2011/01/computers-privacy-and-dataprotection-international-conference-european-data-protection-in-good-healthpart-1/, 2011. [7] U. Altmann, “Representation of Medical Informatics in the Wikipedia and its Perspectives,” Studies in Health Technology and Informatics, vol. 116, pp. 755–760, 2005.

405

406

References

[8] C. Anderson, The Long Tail. How Endless Choice is Creating Unlimited Demand. Croydon: Random House Business Books, 2007. [9] R. M. Anderson and M. M. Funnell, “Patient empowerment: Myths and misconceptions,” Patient Educ Couns, vol. 79, pp. 277–282, 2009. [10] A. Aronson, “DiagnosisPro: The Ultimate Differential Diagnosis Assistant,” Journal of the American Medical Association, vol. 277, no. 5, p. 426, 1997. [11] K. Arrow, “Uncertainty and the Welfare Economics of Medical Care,” The American Economic Review, vol. LIII, no. 5, pp. 941–973, 1963. [12] K. A. Baggerly and K. R. Coombes, “Deriving chemosensitivity from cell lines: Forensic bioinformatics and reproducible research in high-throughput biology,” Annals of Applied Statistics, vol. 3, pp. 1309–1334, 2009. [13] J. E. Bailey, J. Y. Wan, L. M. Mabry, S. H. Landy, R. A. Pope, T. M. Waters, and M. E. Frisse, “Does health information exchange reduce unnecessary neuroimaging and improve quality of headache care in the emergency department?,” J Gen Intern Med, vol. 28, no. 2, pp. 176–183, 2013. [14] A. Bairoch et al., “The Universal Protein Resource (UniProt),” Nucleic Acids Res., vol. 33, no. Database Issue, pp. D154–D159, 2005. [15] A. M. Bardone-Cone and K. M. Cass, “Investigating the impact of proanorexia websites: A pilot study,” European Eating Disorders Review, vol. 14, no. 4, pp. 256–262, 2006. [16] A. M. Bardone-Cone and K. M. Cass, “What does viewing a pro-anorexia website do? An experimental examination of website exposure and moderating effects,” International Journal of Eating Disorders, vol. 40, no. 6, pp. 537–548, 2007. [17] G. A. Barnett and J. M. Hwang, “The Use of the Internet for Health Information and Social Support: A Content Analysis of Online Breast Cancer Discussion Groups,” in The Internet and Health Care: Theory, Research and Practice, (M. Murero and R. E. Rice, eds.), pp. 233–254, Thousand Oaks, CA: Sage, 2006. [18] G. O. Barnett, J. J. Cimino, J. A. Hupp, and E. P. Hoffer, “DXplain: Experience with Knowledge Acquisition and Program Evaluation,” Proc Annu Symp Comput Appl Med Care, vol. 4, pp. 150–154, November 1987. [19] L. Bauld, “The impact of smokefree legislation in England evidence Review University of Bath,” http://www.dh.gov.uk/prod consum dh/groups/dh digitalassets/documents/digitalasset/dh 124959.pdf, 2011. [20] A. Beckett and A. Nayak, “The reflexive consumer,” Marketing Theory, vol. 8, no. 3, pp. 299–317, 2008. [21] D. B´eland, “Ideas and Social Policy: An Institutionalist Perspective,” Social Policy & Administration, vol. 39, no. 1, pp. 1–18, 2005. [22] R. A. Bell, X. Hu, S. E. Orrange, and R. L. Kravitz, “Lingering questions and doubts: Online information-seeking of support forum members following medical visits,” Patient Education & Counseling, vol. 85, no. 3, pp. 525–528, 2011. [23] F. Belleau, M.-A. Nolin, N. Tourigny, P. Rigault, and J. Morissette, “Bio2RDF: Towards a mashup to build bioinformatics knowledge system,” J Biomed Informatics, vol. 41, pp. 706–716, 2008.

References

407

[24] J. Bensing, “Bridging the gap. The separate worlds of evidence-based medicine and patient-centered medicine,” Patient Education and Counselling, vol. 39, pp. 17–25, 2000. [25] H. Berman, K. Henrick, H. Nakamura, and J. L. Markley, “The worldwide Protein Data Bank (wwPDB): Ensuring a single, uniform archive of PDB data,” Nucleic Acids Res, vol. 35, pp. D301–D303, 2007. [26] T. Berners-Lee, “Linked Data — Design Issues,” Latest update 2009, http://www.w3.org/DesignIssues/LinkedData.html, 2006. [27] T. Berners-Lee, “Long live the Web: A call for continued Open Standards and Neutrality. Scientific American,” http://www.scientificamerican.com/ article.cfm?id=long-live-the-web, 2010. [28] T. Berners-Lee, W. Hall, J. A. Hendler, K. O’Hara, N. Shadbolt, and D. J. Weitzner, “Foundations and Trends ,” Web Science, vol. 1, no. 1, pp. 1–130, 2006. 2006 DOI: 10.1561/1800000001. [29] T. Berners-Lee, J. Hendler, and O. Lassila, “The Semantic Web,” Scientific American Magazine, May 17 2001. [30] C. Bizer, T. Heath, and T. Berners-Lee, “Linked Data — The Story So Far,” International Journal on Semantic Web and Information Systems, IJSWIS, vol. 5, no. 3, pp. 1–22, 2009. [31] A. D. Black, J. Car, C. Pagliari, C. Anandan, K. Cresswell, T. Bokun, B. McKinstry, R. Proctor, A. Majeed, and A. Sheikh, “The impact of eHealth on the quality and safety of Healthcare: A systematic overview,” PLoS Med, vol. 8, no. 1, p. e1000387, 2011. Doi:10.1371/journal.pmed1000387. [32] O. Bodenreider, “Biomedical Ontologies in Action: Role in Knowledge Management, Data Integration and Decision Support,” Yearb Med Inform., pp. 67– 79, 2008. [33] A. N. Bonnie, D. J. Schiano, M. Gumbrecht, and L. Swartz, “Why we blog,” Commun. ACM, vol. 47, no. 12, pp. 41–46, December 2004. [34] D. Borzekowski, S. Schenk, J. Wilson, and R. Peebles, “e-Ana and e-Mia: A Content Analysis of Pro–Eating Disorder Web Sites,” American Journal of Public Health, vol. 100, no. 8, pp. 1526–1534, 2010. [35] M. N. Boulos, I. Maramba, and S. Wheeler, “Wikis, blogs and podcasts. a new generation of Web-based tools for virtual collaborative clinical practice and education,” BMC Medical Education, vol. 6, p. 41, 2006. [36] E. A. Bruford, M. J. Lush, M. W. Wright, T. P. Sneddon, S. Povey, and E. Birney, “The HGNC Database in 2008: A resource for the human genome,” Nucleic Acids Res, vol. 36, pp. D445–D448, 2008. [37] M. S. Burgess, O. Allam, W. A. Gray, and A. Al-Busaidi, “Patient-Centred Approach to Focusing Online Health Information Search Results,” in Proc. Int. Symposium on Electronic Personal Health Records 2006 (ISePHR06), pp. 5–16, Sep 28, 2006, Trondheim: Norway, 2006. [38] R. Calvanio, F. S. Buonanno, D. N. Levine, and M. Levine, “Neuropsychiatric sequelae and life events: analysis and management,” in World Stronke Congress, 2008.

408

References

[39] N. C. Campbell, E. Murray, J. Darbyshire, J. Emery, A. Farmer, and F. Griffiths et al., “Designing and evaluating complex interventions to improve health care,” BMJ, vol. 334, no. 7591, pp. 455–459, March 2007. [40] N. Carr, The Shallows. New York: WW Norton and Company, Inc., 2010. [41] K.-H. Cheung, V. Kashyap, J. S. Luciano, H. Chen, Y. Wang, and S. Stephens, “Guest Editorial: Semantic mashup of biomedical data,” Journal of Biomedical Informatics, vol. 41, no. 5, pp. 683–686, October 2008. DOI=10.1016/j.jbi.2008.08.003 http://dx.doi.org/10.1016/j.jbi.2008.08.003. [42] A. Chu and B. Mastel-Smith, “The outcomes of anxiety, confidence, and selfefficacy with internet health information retrieval in older adults: A pilot study,” Computers, Informatics, & Nursing, vol. 28, no. 4, pp. 222–228, 2010. [43] A. Ciolilli, “Phantom authority, self-selective recruitment and retention of members in virtual communities: The case of wikipedia. First Monday, 8, October,” 2003. [44] L. Cohen and L. Manion, Research Methods in Education. London: Routledge Publishers, 4th Edition, 1994. [45] A. Coulter and A. Collins, “Kings Fund: 2011, Making shared decision-making a reality. No decision about me without me,” http://www.kingsfund.org.uk/ publications/making-shared-decision-making-reality, 2011. [46] J. W. Creswell, A. Klassen, V. L. P. Clark, and K. Smith, “Best practices for mixed methods research in the health sciences. Office of Behavioral and Social Sciences Research,” National Institutes of Health, 2011. [47] G. P. Cumming, H. D. Currie, R. Moncur, and A. J. Lee, “Web-based survey on the effect of digital storytelling on empowering women to seek help for urogenital atrophy,” Menopause International, vol. 16, pp. 51–55, 2010. [48] K. Davis, C. Schoen, S. C. Schoenbaum, M. M. Doty, A. L. Holmgren, and J. L. Kriss et al., “Mirror, mirror on the wall: an international update on the comparative performance of American health care. Commonwealth Fund, 2007,” http://www.commonwealthfund.org/Content/Publications/Fund-Reports/ 2007/May/Mirror–Mirror-on-the-Wall–An-International-Update-on-theComparative-Performance-of-American-Healt.aspx Retrieved 31 Jan 2011, 2011. [49] D. deBronkart, “Let Patients Help! A patient engagement handbook How doctors, nurses, patients and caregivers can partner for better care,” March 2013. [50] E. Deci and R. Ryan, Handbook of Self-determination Research. University of Rochester Press, 2002. [51] T. Delbanco, J. Walker, J. D. Darer, J. G. Elmore, H. J. Feldman, S. G. Leveille, J. D. Ralston, S. E. Ross, E. Vodicka, and V. D. Weber, “Open Notes: Doctors and Patients Signing On,” Annals of Internal Medicine, vol. 153, no. 2, pp. 121–125, July 2010. http://annals.org/article.aspx?articleid=745909 Project Website: http://www.myopennotes.org. [52] M. Denscombe, The Good Research Guide. Buckingham: Open University Press, 1998. [53] M. Denscombe, Ground Rules for Good Research: A 10 Point Guide for Social Researchers. Buckingham, UK: Open University Press, 2002.

References

409

[54] Department of Health Equity and Excellence: Liberating the NHS: August 2010, “http://www.dh.gov.uk/en/Publicationsandstatistics/Publications/ PublicationsPolicyAndGuidance/DH 118602,” 2010. [55] Department of Health. Whole system demonstrator programme: Headline Findings: December 2011, “http://www.dh.gov.uk/dr consum dh/groups/ dh digitalassets/documents/digitalasset/dh 131689.pdf,” 2011. [56] J. A. Dewar, “The information age and the printing press: looking backward to see ahead,” http://www.rand.org/content/dam/rand/pubs/papers/ 2005/P8014.pdf, 2005. [57] DiagnosMD. Software DiagnosMD, “Url: http://www.diagnosmd.com/index. php. Last accessed: January, 26,” 2012. [58] J. A. Diaz, C. N. Sciamanna, E. Evangelou, M. J. Stamp, and T. Ferguson, “Brief report: What types of internet guidance do patients want from their physicians?,” Journal of General Internal Medicine, vol. 20, no. 8, pp. 683–685, 2005. [59] M. J. Dutta-Bergman, “Health attitudes, health cognitions, and health behaviors among internet health information seekers: Population-based survey,” Journal of Medical Internet Research, vol. 6, no. 2, p. e15, 2004. [60] Economist, “http://www.economist.com/news/europe/21566664-danishgovernment-rescinds-its-unwieldy-fat-tax-fat-chance,” 2012. [61] G. Elwyn, “Salzburg statement on shared decision making,” BMJ, vol. 342, p. d1745, 2011. [62] T. R. Eng, A. Maxfield, K. Patrick, M. J. Deering, S. C. Ratzan, and D. H. Gustafson, “Access to Health Information and Support: A Public Highway or a Private Road?,” JAMA, vol. 280, no. 15, pp. 1371–1375, 1998. [63] R. M. Epstein, B. S. Alper, and T. E. Quill, “Communicating evidence for participatory decision making,” JAMA, vol. 291, no. 19, pp. 2359–2366, 2004. [64] G. Eysenbach, “What is eHealth?,” J Med Internet Res, vol. 3, no. 2, p. e20, 2001. URL: http://www.jmir.org/2001/2/e20/, doi: http://dx.doi.org/ 10.2196/ jmir.3.2.e20 PMID: 11720962. [65] G. Eysenbach, “Infodemiology: The epidemiology of (Mis) information,” American Journal of Medicine, vol. 113, pp. 763–765, 2002. [66] G. Eysenbach, “The impact of the Internet on cancer outcomes,” CA Cancer J Clin., vol. 53, pp. 356–371, 2003. [67] G. Eysenbach, “Medicine 2.0: Social Networking, Collaboration, Participation, Apomediation, and Openness,” J Med Internet Res, vol. 10, no. 3, p. e22, 2008. URL: http://www.jmir.org/2008/3/e22/ doi:10.2196/jmir.1030. [68] G. Eysenbach, J. Powell, M. Englesakis, C. Rizo, and A. Stern, “Health related virtual communities and electronic support groups: systematic review of the effects of online peer-to peer-interactions,” BMJ, vol. 328, p. 1166, 2004. [69] R. Fawdry, S. Bewley, G. Cumming, and H. Perry, “Paradigm shift in maternity IT?,” Journal of the Royal Society of Medicine, vol. 104, no. 10, pp. 405–412, 2011. [70] M. Y. Feilzer, “Doing Mixed Methods Research Pragmatically: Implications for the rediscovery of pragmatism as a research paradigm,” J Mixed Methods Research, vol. 4, no. 1, pp. 6–16, 2010.

410

References

[71] B. J. Fogg, Persuasive Technology: Using Computers to Change What We Think and Do. San Francisco: Morgan Kaufmann, 2003. [72] S. Fox, “Health Topics, February 1, 2011, Pew Internet Project and the California HealthCare Foundation,” http://www.pewinternet.org/∼/media// Files/Reports/2011/PIP Health Topics.pdf, 2011. [73] S. Fox and R. Lee, “The online healthcare revolution: How the web helps Americans take better care of themselves. The Pew Internet and American Life Project,” Washington DC, 2000. [74] S. Fox and K. Purcell, “Pew Internet and American Life Project 2010 Chronic disease and the Internet http://www.pewinternet.org/Reports/2010/ChronicDisease.aspx,” 2010. [75] M. Fraser and S. Dutta, Throwing Sheep in the Boardroom: How Online Social Networking Will Transform Your Life. Work and World. Hoboken, NJ: Wiley, 2008. [76] P. Freire, Pedagogy of the Oppressed. New York NY: Continuum Press, 1970. [77] M. E. Frisse, K. B. Johnson, H. Nian, C. L. Davison, C. S. Gadd, K. M. Unertl, P. A. Turri, and Q. Chen, “The financial impact of health information exchange on emergency department care,” J Am Med Inform Assoc, vol. 19, no. 3, pp. 328–333, 2011. [78] J. H. Frost and M. P. Massagli, “Social uses of personal health information within PatientsLikeMe, an online patient community: what can happen when patients have access to one another’s data,” J Med Internet Res., vol. 10, no. 3, p. e15, 2008. http://www.ncbi.nlm.nih.gov/pmc/articles/pmc2553248/. [79] A. Gagliardi and A. Jadad, “Examination of instruments used to rate quality of health information on the internet; chronicle of a voyage with an unclear destination,” BMJ, vol. 324, pp. 569–572, 2002. [80] J. Gallucci, Not Pro-Ana: The Online Sisterhood of Eating Disorders. Long Island Press, 2010. retrieved 10 December. [81] V. Gewin, “Research: Uncovering misconduct,” Nature, vol. 485, pp. 137–139, 2012. [82] M. Gill and R. Stott, “Health professionals must act to tackle climate change,” The Lancet, vol. 374, no. 9706, pp. 1953–1955, 2009. [83] M. Gillam et al., “The Healthcare Singularity and the Age of Semantic Medicine,” in The Fourth Paradigm: Data-Intensive Scientific Discovery, (T. Hey, ed.), 2009. [84] M. J. K. Gloria, J. S. Luciano, and D. L. McGuinness, “Building SemanticallyEnriched Web Observatories,” in WWW ’13 Companion: International World Wide Web Conference, 2013. [85] C. A. Goble, J. Bhagat, S. Aleksejevs, D. Cruickshank, D. Michaelides, D. Newman, and M. Borkum et al., “myExperiment: A repository and social network for the sharing of bioinformatics workflows,” Nucleic Acids Research, vol. 38, no. Web Server issue, pp. W677–682, 2010. [86] C. A. Goble and R. Stevens, “The state of the nation in data integration,” J Biomed Informatics, vol. 41, pp. 687–693, 2008. [87] B. Goldacre, Bad Pharma: How Drug Companies Mislead Doctors and Harm Patients. London: Fourth Estate, 2012.

References

411

[88] M. L. Graber and A. Mathew, “Performance of a Web-Based Clinical Diagnosis Support System for Internists,” Journal of General Internal Medicine, vol. 23, no. Supplement 1, pp. 37–40, 2008. [89] T. Greenhalgh, “Why do we always end up here? Medicine’s conceptual cul de sacs and some off-road alternative routes,” J. Prim. Health Care, vol. 4, no. 2, pp. 92–97, 2012. [90] T. Greenhalgh, S. Hinder, K. Stramer, T. Bratan, and J. Russel, “Adoption, non-adoption, and abandonment of a personal electronic health record: Case study of HealthSpace,” BMJ, vol. 341, p. c5814, 2010. [91] T. Greenhalgh and J. Russell, “Why do evaluations of eHealth programs fail? An alternative set of guiding principles,” Plos Med, vol. 7, no. 11, p. e1000360, 2010. doi10.1371journal.pmed.1000360. [92] T. Greenhalgh, K. Stramer, T. Bratan, E. Byrne, J. Russell, S. Hinder, and H. Potts, “The Devil’s in the Detail: Final report of the independent evaluation of the Summary Care Record and HealthSpace programmes,” London: University College London, 2010, https://www.ucl.ac.uk/news/scriefullreport.pdf. [93] S. L. Greer, “The Changing World of European Health Lobbies,” in Lobbying the European Union: Institutions, Actors, and Issues, (D. Coen and J. Richardson, eds.), Oxford: Oxford University Press, 2009. [94] F. Griffiths, A. Lindenmeyer, J. Powell, P. Lowe, and M. Thorogood, “Why are health care interventions delivered over the Internet? A systematic review of the published literature,” J Med Internet Res, vol. 8, no. 2, p. e10, 2006. http://www.jmir.org/2006/2/e10/. [95] L. Grossman, “Time’s Person of the year: You,” http://www.time.com/ time/magazine/article/0,9171,1569514,00.html, 2006. [96] E. M. Guizzo, “The Essential Message: Claude Shannon and the Making of Information Theory,” http://dspace.mit.edu/bitstream/handle/1721.1/ 39429/54526133.pdf. [97] M. Gymrek, A. L. McGuire, D. Golan, E. Halperin, and Y. Erlich, “Identifying personal genomes by surname inference,” Science, vol. 339, no. 6117, pp. 321– 324, January 2013. [DOI:10.1126/science.1229566]. [98] P. A. Hall, “The Role of Interests, Institutions, and Idea in the Comparative Political Economy of the Industrialized Nations,” in Comparative Politics: Rationality, Culture, and Structure, (M. I. Lichbach and A. S. Zuckerman, eds.), pp. 174–208, Cambridge: Cambridge University Press, 1997. [99] P. A. Hall and R. Taylor, “Political Science and the Three New Institutionalisms,” Political Studies, vol. XLIV, pp. 936–957, 1996. [100] K. Hampton, L. Sessions Goulet, C. Marlow, and L. Rainie, “Why Most Facebook Users Get More Than They Give. Pew Internet and American Life Project: Social Networking Reports,” Available at: http://www.pewinternet.org/ Reports/2012/Facebook-users.aspx, 3 February 2012. [101] J. C. R. Hardwick and F. M. MacKenzie, “Information contained in miscarriage-related websites and the predictive value of website scoring systems,” Eur J Obstet Gynecol Reprod Biol, vol. 106, pp. 60–63, 2003.

412

References

[102] J. Harshbarger, C. Ahlers-Schmidt, L. Mayans, D. Mayans, and J. Hawkins, “Pro-anorexia websites: What a clinician should know,” International Journal of Eating Disorders, vol. 42, no. 4, pp. 367–370, 2008. [103] M. Hauder et al., “Making data analysis expertise broadly accessible through workflows,” in Proc. WORKS ’11, p. 77, New York, USA: ACM Press, 2011. [104] B. Hauray, L’Europe du m´edicament. Politique — Expertise — Int´erˆets Priv´es. Paris: Presses de Sciences Po, 2006. [105] Health and Environmental Alliance, “ProtectingPublic Health From Climate Change – A “Global” Call “to” Action,” http://www.env-health.org/IMG/ pdf/Durban Global Climate and Health Call to Action Final.pdf, 2011. [106] J. P. Heinz, E. O. Laumann, R. L. Nelson, and R. H. Salisbury, The Hollow Core: Private Interests in National Policymaking. Cambridge: Harvard University Press, 1993. [107] J. Hendler, N. Shadbolt, W. Hall, T. Berners-Lee, and D. Weitzner, “Web Science: An interdisciplinary approach to understanding the web,” Communication of the ACM, vol. 51, no. 7, pp. 60–69, 2008. [108] B. W. Hesse, “Harnessing the power of an intelligent health environment in cancer control,” Studies in Health Technology & Informatics, vol. 118, pp. 159– 176, 2005. [109] H.Masum and R. Harris, Open Source for Neglected Diseases: Magic Bullet or Mirage? Washington, DC: Results for Development Institute, 2011. [110] Internet World Stats, “http://www.internetworldstats.com/stats.htm,” Miniwatts Marketing Group, 2012, Page updated 28 March. [111] J. P. A. Ioannidis et al., “Repeatability of published microarray gene expression analyses,” Nature Genetics, vol. 41, pp. 149–155, 2009. [112] H. J, “Pew Internet and American Life Project 2009,” The mobile difference http://www.pewinternet.org/Reports/2009/5-The-Mobile-Difference– Typology.aspx, 2009. [113] A. R. Jadad, M. W. Enkin, S. Glouberman, P. Groff, and A. Stern, “Are virtual communities good for our health?,” BMJ, vol. 332, no. 7547, pp. 939– 942, 2006. [114] H. Jarman and S. Greer, “Crossborder Trade in Health Services: Lessons from the European Laboratory,” Health Policy, vol. 94, pp. 158–163, 2010. [115] A. Jentzsch, J. Zhao, O. Hassanzadeh, K. Cheung, K. Samwald, and B. Andersson, “Linking open drug data,” in Proceedings of the International Conference on Semantic Systems (I-SEMANTICS’09), Graz, Austria. http://www.jcheminf.com/content/3/1/19, 2009. [116] S. Jett, D. J. LaPorte, and J. Wanchisn, “Impact of exposure to pro-eating disorder websites on eating behaviorin college women,” European Eating Disorders Review, vol. 18, no. 5, pp. 410–416, 2010. [117] E. Kahana and B. Kahana, “Patient Proactivity Enhancing Doctor-PatientFamily Communication in Cancer Prevention and Care Among the Aged,” Patient Education and Counselling, vol. 2075, pp. 1–7, 2003. [118] E. Kahana and B. Kahana, “Health Care Partnership Model of doctor-patient communication in cancer prevention and care among the aged,” in Handbook of Communication and Cancer Care, (D. O’Hair, G. Kreps, and L. Sparks, eds.), pp. 37–54, New York: Hampton Press, 2007.

References

413

[119] D. Kahneman, Thinking, Fast and Slow. Farrar, Straus and Giroux, 2011. [120] M. Kanehisa and S. Goto, “KEGG: Kyoto Encyclopedia of Genes and Genomes,” Nucleic Acids Res, vol. 28, pp. 27–30, 2000. [121] B. Kaplan and K. D. Harris-Salamone, “Health IT Success and Failure: Recommendations from Literature and an AMIA Workshop,” J Am Med Inform Assoc, vol. 16, no. 3, pp. 291–299, May–June 2009. [122] R. Keeney, “Personal Decisions are the leading cause of death,” Operations Research, vol. 56, pp. 1335–1347, 2008. [123] N. Kobie, “Q&A: Conrad Wolfram on communicating with apps in Web 3.0. IT Pro,” Available at http://www.itpro.co.uk/621535/qa-conrad-wolfram-oncommunicating-with-apps-in-web-30, March 17 2010. [124] E. Kolber, “Enter the Anthropocene — Age of Man, National Geographic,” http://ngm.nationalgeographic.com/2011/03/age-of-man/kolberttext, March 2011. [125] R. Kurzweil, “The Law of accelerating returns,” http://www.kurzweilai.net/ the-law-of-accelerating-returns, 2011. [126] A. Lang and A. Mertes, “eHealth Policy and Deployment Activities in Europe,” vol. 17, no. 4, pp. 262–268, 2011. [127] B. LeBlanc, “An Exploratory Study of ‘Bug Chasers’,” Sociological Imagination, vol. 43, no. 2, pp. 13–20, 2007. [128] R. Lenz, M. Peleg, and M. Reichert, “Healthcare Process Support: Achievements, Challenges, Current Research,” International Journal of KnowledgeBased Organizations (IJKBO), vol. 2, no. 4, 2012. [129] M. Levine, R. J. Calvanio, and F. S. Buonnano, “How tracking neuropsychiatric symptom expression can improve diagnosis and management,” Presented at Medicine 2.0, September, 2012. [130] M. Levine and G. Mesibov, “Personal guidance system for school and vocational functioning,” in Poster presented at the Annual International Meeting for Autism Researchers, Seattle, WA, May 3–5 2007. [131] G. H. Lewes, Problems of Life and Mind. Vol. 2, London: Tr¨ ubner, First series Edition, 1879. [132] R. C. Lieberman, “Ideas, Institutions, and Political Order: Explaining Political Change,” American Political Science Review, vol. 96, no. 4, pp. 697–712, 2002. [133] R. J. Lilford, J. Foster, and M. Pringle, “Evaluating eHealth: How to make evaluation more methodologically robust,” Plos Med, vol. 6, no. 11, p. e1000186, 2009. Doi:10.1371/journal.pmed.1000186. [134] M. J. Lincoln, C. W. Turner, P. J. Haug, H. R. Warner, J. W. Williamson, O. Bouhaddou, S. G. Jessen, D. Sorenson, R. C. Cundick, and M. Grant, “Iliad training enhances medical students’ diagnostic skills,” Journal of Medical Systems, vol. 15, no. 1, pp. 93–110, 1991. [135] Littlejohns et al., “Evaluating computerized health information systems: Hard lessons still to be learnt,” BMJ, vol. 326, pp. 860–863, 2003. [136] D. Lucci, M. Levine, K. Challen, and D. S. McLeod, “Technologies to Support Interventions for Social Emotional Intelligence, Self-Awareness/Personal Style, and Self-Regulation,” in Handbook of Learning Technologies for People with Autism Spectrum and Related Disorders, Chapter 10, (K. B. M. Goodwin and S. Wayland, eds.), MD: Paul H. Brookes Pub, 2013.

414

References

[137] D. Lucci, D. S. McLeod, and K. Challen-Wittmer, “Male teens with Asperger’s syndrome and nonverbal learning disorder learn about stress and its physiological signs,” in Poster Presented at the Annual International Meeting for Autism Research, Philadelphia, PA, May 20–22 2010. [138] B. Lud¨ ascher, I. Altintas, C. Berkley, D. Higgins, E. Jaeger-Frank, M. Jones, E. Lee, J. Tao, and Y. Zhao, “Scientific Workflow Management and the Kepler System,” Special Issue: Workflow in Grid Systems. Concurrency and Computation: Practice & Experience, vol. 18, no. 10, pp. 1039–1065, 2006. [139] G. Makoul and R. H. Curry, “The value of assessing and addressing communication skills,” Journal of the American Medical Association, vol. 298, no. 9, pp. 1057–1059, 2007. [140] H. Margetts, “The Internet and Public Policy,” Policy & Internet, vol. 1, no. 1, pp. 1–21, 2009. [141] J. McCusker, D. L. McGuinness, J. Lee, C. Thomas, P. Courtney, Z. Tatalovich, N. Contractor, G. D. Morgan, and A. R. Shaikh, “Towards Next Generation Health Data Exploration: A Data Cube-based Investigation into Population Statistics for Tobacco,” in Proceedings of Hawaii International Conference on System Sciences 2013, Grand Wailea, Maui, Hawaii, January 7–10 2013. [142] M. McLuhan, Understanding Media: The Extensions of Man. McGraw Hill, NY, 1964; reissued MIT Press, 1994: reissued by Gingko Press, 1st Edition, 2003. [143] Medicine 2.0 Congress 2012 http://www.medicine20congress.com/ocs/index. php/med/med2012, 2012. [144] P. E. Meehl, “Clinical versus statistical prediction: A theoretical analysis and a review of the evidence,” 1954. [145] R. A. Miller, H. E. Pople, and J. D. Myers, “Internist-I, an Experimental Computer-Based Diagnostic Consultant for General Internal Medicine,” N Engl J Med, vol. 307, no. 8, pp. 468–476, August 1982. [146] A. Moraru and D. Mladenic, “A framework for semantic enrichment of sensor data,” CIT Journal of Computing and Information Technology, vol. 20, no. 3, p. 167, September 2012. [147] L. Moreau, “The Foundations for Provenance on the Web,” Foundations and Trends in Web Science, vol. 2, no. 2–3, pp. 99–241, 2010. [DOI: 10.1561/1800000010]. [148] D. L. Morgan, “Paradigms lost and pragmatism regained. Methodological Implications of combining qualitatative and quantitative methods,” Journal of Mixed Methods Research, vol. 1, pp. 48–76, 2007. [149] M. Morris, E. Ozanne, K. Miller, N. Santamaria, A. Pearce, C. Said, and B. Adair, “Smart technologies for older people: A systematic literature review of smart technologies that promote health and well-beingwell-beingwell-being of older people living at home Institute for a Broadband-Enabled Society, Level 4, Building 193. The University of Melbourne, Victoria 3010, Australia. ISBN 978 0 7340 4781 6,” http://www.webcitation.org/6CwGZwwz2, 2012.

References

415

[150] Z. S. Morris, S. Wooding, and J. Grant, “The answer is 17 years, what is the question: Understanding time lags in translational research,” JSRM, vol. 104, pp. 510–520, 2011. [151] D. A. Moskowitz and M. E. Roloff, “The existence of a bug chasing subculture,” Culture, Health & Sexuality, vol. 9, no. 4, no. 4, pp. 347–358, 2007. [152] D. A. Moskowitz and M. E. Roloff, “The ultimate high: Sexual addiction and the bug chasing phenomenon,” Sexual Addiction and Compulsivity, vol. 14, no. 1, no. 1, pp. 21–40, 2007. [153] E. Murray, “eHealth for self-management and behaviorchange: Potential, progress and pitfalls,” in Medicine 2.0 Conference, Maastricht 2010, Retrieved 31 Jan 2011. http://www.medicine20congress.com/ocs/ index.php/med/med2010/paper/view/482. [154] E. Murray, J. Burns, and S. See Tai et al., “Interactive health communication applications for people with chronic disease,” Cochrane Database Syst Rev, vol. 4, p. CD004274, 2005. [155] E. Murray, S. Treweek, C. Pope, A. MacFarlane, L. Ballini, and C. Dowrick et al., “Normalisation process theory: A framework for developing, evaluating and implementing complex interventions,” BMC Medicine, vol. 8, p. 63, October 2010. http://www.biomedcentral.com/1741-7015/8/63 Retrieved 31 Jan 2011. [156] V. A. Narayan, M. Mohwinckel, G. Pisano, M. Yang, and H. K. Manji, “Beyond magic bullets: True innovation in health care,” Nature Reviews Drug Discovery, vol. 12, pp. 85–86, 2013. [157] H. Q. Nguyen and D. Donestky-Cuenco et al., “Randomized control trial of an Internet-based versus face-to-face dypsnea self-management program for patients with chronic obstructive pulmonary disease pilot study,” J. Med. Internet Res, vol. 10, no. 2, no. 2, p. e9, 2008. [158] NHS Institute for Innovation and Improvement, “Patient Experience Network (PEN),” http://www.institute.nhs.uk/share and network/pen/coproduction.html. [159] E. Nolte and C. M. McKee, “In Amenable Mortality — Deaths Avoidable Through Health Care — Progress in the US Lags That of Three European Countries,” Health Affairs Web First, published online Aug. 29, 2012. [160] M. L. Norris, K. M. Boydell, L. Pinhas, and D. K. Katzman, “Ana and the internet: A review of pro-anorexia websites,” The International Journal of Eating Disorders, vol. 39, no. 6, no. 6, pp. 443–447, 2006. [161] D. North, “Institutions,” Journal of Economic Perspectives, vol. 5, no. 1, no. 1, pp. 97–112, 1991. [162] T. Oinn, M. Addis, J. Ferris, D. Marvin, M. Senger, M. Greenwood, T. Carver, and K. Glover et al., “Taverna: A tool for the composition and enactment of bioinformatics workflows,” Bioinformatics, vol. 20, no. 17, no. 17, pp. 3045– 3054, 2004. [163] T. O’Reilly, “O’Reilly Media,” What is Web 2.0? http://oreilly.com/web2/ archive/what-is-web-20.html, 2005.

416

References

[164] J. Osborne, J. Flatow, M. Holko, S. Lin, W. Kibbe, L. Zhu, M. Danila, G. Feng, and R. Chisholm, “Annotating the humangenome with disease ontology,” BMC Genomics, vol. 10, no. Suppl 1, no. Suppl 1, p. S6, 2009. [165] C. Pagliari, D. Detmer, and P. Singleton, “Potential of electronic personal health records,” BMJ, vol. 335, no. 7615, pp. 330–333, August 2007. [166] E. Pariser, What the Internet is Hiding from You. USA: Penguin Books, 2011. [167] A. Payne, K. Storbacka, and P. Frow, “Managing the co-creation of value,” Journal of the Academy of Marketing Science, vol. 36, no. 1, no. 1, pp. 83–96, 2008. [168] N. Postman, “Five Things We Need to Know About Technological Change,” Speech delivered by Postman March 27, 1998 to a gathering of theologians and religious leaders in Denver, Colorado. http://www.michaelzimmer.org/ 2008/01/04/neil-postman-five-things-we-need-to-know-about-technologicalchange/, 1998. [169] PricewaterhouseCoopers, “Good medicine for public-private healthcare,” https://www.pwc.com/en VN/vn/publications/assets/IGU-good-medicinefor-public-private-healthcare.pdf, 2009. [170] S. Rafaeli and Y. Ariel, “Online motivational factors: Incentives for participation and contribution in wikipedia,” Psychological Aspects of Cyberspace, 23, February 2008. [171] R. Ransom and J. Hepworth, “An interpretative phenomenological analysis of participation in a pro-anorexia internet site and its relationship with disordered eating,” Journal of Health Psychology, vol. 11, no. 2, no. 2, pp. 283–296, 2006. [172] P. Resnick and H. R. Varian, “Recommender systems,” Communications of the ACM (CACM), vol. 40, no. 3, March 1997. doi: 10.1145/245108.245121. [173] A. Risk and J. Dzenowagis, “Review Of Internet Health Information Quality Initiatives,” J Med Internet Res, vol. 3, no. 4, p. e28, 2001. URL: http://www.jmir.org/2001/4/e28/. [174] K. J. Roberts, “Patient empowerment in the United States: A critical commentary,” Health Expectations, vol. 2, no. 2, pp. 1369–7625, 1999. Blackwell Science Ltd. [175] A. Robertson, T. Cornford, N. Barber, T. Avery, and A. Sheikh, “The NHS IT project: More than just a bad dream,” Lancet, vol. 379, no. 9810, pp. 29–30, January 2012. http://www.sciencedirect.com/science/article/pii/ S0140673612600233-aff1. [176] R. Robinson-O’Brien, C. L. Perry, M. Wall, M. Story, and D. NeumarkSztainer, “Adolescent and Young Adult Vegetarianism: Better Dietary Intake and Weight Outcomes but Increased Risk of Disordered Eating Behaviors,” Journal of the American Dietetic Association, vol. 108, no. 4, pp. 648–655, 2009. [177] A. Rodr´ıguez-Gonz´ alez, J. Torres-Ni˜ no, M. A. Mayer, G. Alor-Hernandez, and M. D. Wilkinson, “Analysis of ML-DDSS, a multi-level diagnosis decision support system, and its implications: A case study,” Computational and Mathematical Methods in Medicine, in press.

References

417

[178] P. Roffe, G. Tansey, and D. Vivas-Eugui, eds., Negotiating Health: Intellectual Property and Access to Medicines. London: Earthscan, 2006. [179] H.-W. Romana, “Is evidence-based medicine patient-centered and is patientcentered care evidence-based?,” Health Serv Res, vol. 41, no. 1, pp. 1–8, 2006. [180] Royal College of Obstetricians and Gynaecologists, “Scientific Advisory Committee Opinion Paper No. 27 Why we should consider a life course approach to women’s health,” http://www.rcog.org.uk/files/rcogcorp/14.10.11SACLifecourse.pdf, 2011. [181] A. Ruttenberg, T. Clark, W. Bug, M. Samwald, O. Bodenreider, H. Chen, D. Doherty, K. Forsberg, Y. Gao, V. Kashyap, J. Kinoshita, J. Luciano, M. S. Marshall, C. Ogbuji, J. Rees, S. Stephens, G. T. Wong, E. Wu, D. Zaccagninil, T. Hongsermeier, E. Neumann, I. Herman, and K.-H. Cheung, “Advancing translational research with the Semantic Web,” BMC Bioinformatics, vol. 8, no. Suppl 3, p. S2, 2007. http://www.biomedcentral.com/1471-2105/8/S3/S2. [182] R. B. Saltman, “Patient Choice and Patient Empowerment in Northern European Health Systems: A Conceptual Framework,” International Journal of Health Services, vol. 24, no. 2, pp. 201–229, 1994. DOI: 10.2190/8WMPRR2K-ABM7-NVNH. [183] C. Schoen, R. Osborn, D. Squires, M. M. Doty, P. Rasmussen, R. Pierson, and S. Applebaum, “A Survey of Primary Care Doctors in Ten Countries Shows Progress in Use of Health Information Technology, Less in Other Areas,” Health Affairs Web First, published online Nov. 15, 2012. [184] P. J. Scott and J. S. Briggs, “A pragmatist argument for mixed methodology in medical informatics,” Journal Of Mixed Methods Research, vol. 3, no. 3, pp. 223–241, 2009. [185] Scottish Government, “A National Telehealth and Telecare: Delivery Plan for Scotland to 2015 Driving Improvement,” Integration and Innovation. Produced for the Scottish Government by APS Group Scotland DPPAS 13767 (12/12), December 2012, ISBN: 978-1-78256-317-4, http://www.sctt.scot.nhs.uk/wpcontent/uploads/2012/12/449260 v7 20121221.pdf. [186] C. E. Shannon, “A Mathematical Theory of Communication,” The Bell System Technical Journal, vol. 27, pp. 379–423, 623–656, 1948. http://cm.belllabs.com/cm/ms/what/shannonday/shannon1948.pdf. [187] B. Sibbald and M. Roland, “Understanding controlled trials. Why are randomised controlled trials important?,” BMJ, vol. 316, no. 7126, p. 201, January 1998. [188] M. Slater, “Craving Community: The Phenomenon of Pro-Anorexia Sites,” AngeLingo (USC College of Letters Arts and Sciences), vol. 4, no. 2, 2006. [189] A. Smith and J. Brenner, “Twitter Use 2012,” Pew Internet and American Life Project: Social Networking Reports, May 31st, Available at http://www.pewinternet.org/Reports/2012/Twitter-Use-2012.aspx, 2012. [190] G. C. S. Smith and J. P. Pell, “Parachute use to prevent death and major trauma related to gravitational challenge: Systematic review of randomised controlled trials,” BMJ, December 2003.

418

References

[191] P. C. Smith, A. Stepan, V. Valdmanis, and P. Verheyen, “Principal-agent problems in health care systems: An international perspective,” Health Policy, vol. 41, no. 1, pp. 37–60, 1997. [192] R. Smith BMJ group blogs http://blogs.bmj.com/bmj/2011/11/22/richardsmith-can-information-technology-improve-healthcare/, 2011. [193] R. Smith, “http://blogs.bmj.com/bmj/2011/11/22/richard-smith-caninformation-technology-improve-healthcare/,” 2011. [194] M. R. Somers and F. Block, “From Poverty to Perversity: Ideas, Markets, and Institutions over 200 Years of Welfare Debate,” American Sociological Review, vol. 70, no. 2, pp. 260–287, 2005. [195] Steventon et al., “Effect of telehealth on use of secondary care and mortality: Findings from the Whole System Demonstrator cluster randomized trial,” BMJ, vol. 344, p. e3874, 2012. [196] M. A. Stewart, “Effective physician-patient communication and health outcomes: A review,” Canadian Medical Association, vol. 152, no. 9, pp. 1423– 1433, 1995. [197] G. J. Stigler, “The Economics of Information,” Journal of Political Economy, vol. 69, no. 3, pp. 213–225, 1961. [198] K. G. Sweeney, D. MacAuley, and D. P. Gray, “Personal significance: The third dimension,” Lancet, vol. 351, pp. 134–136, 1998. [199] H. Tang and J. H. K. Ng, “Googling for a diagnosis — use of Google as a diagnostic aid: Internet based study,” BMJ, vol. 333, no. 7579, p. 1143, 2006. [200] L. E. Tesler and R. E. Malone, “Corporate Philanthropy, Lobbying, and Public Health Policy,” American Journal of Public Health, vol. 98, no. 12, pp. 2123– 2133, 2008. [201] The Burrill Report, “A Global Health Care Crisis,” http://www.burrillreport.com/ article-a global healthcare crisis.html, 2010. [202] The NCD alliance 2012, “Putting non-communicable diseases on the global agenda,” http://ncdalliance.org/sites/default/files/resource files/ NCDA EMRORC.pdf, 2012. [203] A. Tiwaria and K. T. Arvind, “Sekhar: Workflow based framework for life science informatics,” Computational Biology and Chemistry, vol. 31, no. 5–6, pp. 305–319, October 2007. [204] T. H. Van De Belt, L. J. Engelen, S. A. A. Berben, and L. Schoonhoven, “Definition of Health 2.0 and Medicine 2.0: A Systematic Review,” J Med Internet Res, vol. 12, no. 2, p. e18, 2010. URL: http://www.jmir.org/2010/2/e18/ndoi: 10.2196/jmir.1350 PMID: 20542857. [205] M. H. Van den Berg and J. K. Ronday et al., “Using Internet technology to deliver a home-based physical activity intervention for patients with Rheumatoid Arthritis: A randomized control trial,” Arthritis & Rheumatism, vol. 55, no. 6, pp. 935–945, 2006. [206] C. F. van Uden-Kraan, C. H. C. Drossaert, and E. Taal et al., “Participation in online patient support groups endorses patients’ empowerment,” Patient Educ Couns, vol. 74, pp. 61–69, 2009.

References

419

[207] B. Vandervalk, L. McCarthy, and M. Wilkinson, “SHARE: A Semantic Web Query Engine for Bioinformatics, vol. 5926,” in The Semantic Web: Proceedings of the ASWC, pp. 367–369, 2009. [208] N. Villanueva-Rosales and M. Dumontier, “yOWL: An ontology-driven knowledge base for yeast biologists,” J Biomed Informatics, vol. 41, pp. 779–789, 2008. [209] J. Walker, Mobilizing Interest Groups in America. Ann Arbor: University of Michigan Press, 1991. [210] A. Wall, “The Atlantic Online. Search Engine History,” http://www.searchenginehistory.com/, 2012. [211] T. L. Webb, J. Joseph, L. Yardley, and S. Michie, “Using the Internet to promote health behavior change: A systematic review and meta-analysis of the impact of theoretical basis, use of behavior change techniques, and mode of delivery on efficacy,” J Med Internet Res, vol. 12, no. 1, p. e4, 2010. [212] WHO, “Chronic diseases and health promotion, http://www.who.int/ chp/en/,” 2013. [213] WHO Global Observatory for eHealth, Telemedicine: Opportunities and Developments for Member States. Geneva: World Health Organization, 2010. [214] WHO/Europe, Empowering patients. (n.d.), “WHO/Europe, World Health Organization Regional Office for Europe,” from http:// www.euro.who.int/en/what-we-do/health-topics/noncommunicable-diseases/ sections/news/2012/4/empowering-patients, Retrieved December 15 2012. [215] P. Wilson, “How to find the good and avoid the bad or ugly: A short guide to tools for rating quality of health information on the internet,” BMJ, vol. 324, pp. 598–600, 2002. [216] D. S. Wishart et al., “DrugBank: a comprehensive resource for in silico drug discovery and exploration,” Nucleic Acids Res., vol. 43, pp. D668–D672, 2006. [217] I. Wood and B. V. et al, “OWL-DL domain-models as abstract workflows,” ISoLA 2, vol. 7610, pp. 56–66, 2012. [218] D. Wooton, Bad Medicine: Doctors Doing Harm Since Socrates. Oxford University Press, 2006. [219] World Health Organization, Equitable Access to Essential Medicines: A Framework for Collective Action. Geneva: WHO Policy Perspectives on Medicines, 2004. [220] M. K. Wynia, “Public health, public trust, and lobbying,” American Journal of Bioethics, vol. 7, no. 6, pp. 4–7, 2007. [221] C. Xie, R. Bagozzi, and S. Troye, “Trying to prosume: Toward a theory of consumers as co-creators of value,” Journal of the Academy of Marketing Science, vol. 36, no. 1, pp. 109–122, 2008. [222] V. L. Yua, B. G. Buchananb, E. H. Shortliffe, S. M. Wraith, R. Davis, A. C. Scott, and S. N. Cohen, “Evaluating the performance of a computer-based consultant,” Computer Programs in Biomedicine, vol. 9, no. 1, pp. 95–102, January 1979.