On October 29-30, 2009, 97 people gathered at the Parc 55 Hotel in San Francisco, CA, for the “Discovery and Innovation in Health IT” workshop (see http://www.cra.org/ccc/healthit.php). This invitation-only event, co-sponsored by several federal agencies and non-profit organiza-tions, sought—through a series of plenary and breakout sessions—to explore and define fun-damental research challenges and opportunities in using information technology to improve health and healthcare.
In co-sponsoring the workshop, the National Science Foundation (NSF), Office of the National Coordinator for Health Information Technology (ONC), National Institute of Standards and Technology (NIST), National Library of Medicine (NLM), Agency for Healthcare Research and Quality (AHRQ), Computing Community Consortium (CCC), and American Medical Infor-matics Association (AMIA) asked leading computer scientists, medical practitioners, systems engineers, and social scientists spanning academia, industry, and government to identify mutual interests in health IT, as they relate to near- and long-term challenges and solutions; and to de-fine a range of “model” proof-of-concept, integrative systems that might serve as motivating and unifying forces to drive fundamental research in health IT and accelerate the translation of re-search outcomes into products and services.
The workshop was structured as a series of four half-day sessions. The first three ses-sions were comprised of two plenary talks followed by small-group (about 10 people per group) breakout discussions in which participants defined particular research challenges, multiple lines of attack, and possible test-beds or demonstration systems. At the conclusion of each session, the groups delivered short presentations summarizing their conversations. (Videos of the ple-nary talks and report-outs will be available at www.cra.org/ccc/healthit.php shortly.) Workshop organizers selected the group participants for the first two sessions, opting to assemble people with diverse backgrounds within each group, while participants self-selected the groups for the third session. The groups were asked to approach their discussions by considering the “per-spectives” of patients and caregivers during the first session; “processes” such as prevention, prediction, diagnosis, intervention, rehabilitation, and monitoring, etc., during the second ses-sion; and their own research interests during the third session. The fourth half-day session pro-vided an opportunity for workshop participants to synthesize the research opportunities defined throughout the earlier parts of the meeting and frame a call-to-action for the future.
The workshop was comprised of individuals carefully selected to represent the wide va-riety of interests, expertise areas, needs, and constituencies within the healthcare arena. Their backgrounds spanned robotics, imaging, mobility and sensing, decision support systems, elec-tronic health records, privacy and security, genomics, data mining and analysis, social and be-havioral science, and human-computer interaction.
These diverse research interests were apparent during the workshop’s plenary talks. William Stead, M.D., and Latanya Sweeney, Ph.D., led off the meeting. Stead, the Chief Strat-egy and Information Officer as well as a Professor of Medicine at Vanderbilt University, summa-rized the results and recommendations of a National Research Council report on “Computa-tional Technology for Effective Healthcare: Immediate Steps and Strategic Directions” that he co-edited earlier this year. Stead illustrated key research challenges, including patient-centered cognitive support, modeling, automation, data sharing and collaboration, data management at scale, and automated capture of patient-doctor interactions. Sweeney, Distinguished Career Professor of Computer Science, Technology, and Policy at Carnegie Mellon University’s School of Computer Science, and a member of the Federal Health Information Technology Policy Committee, described research problems for computer science and information technology in the context of a national health information infrastructure. In particular, she articulated steps to data consolidation, analytics, and privacy/confidentiality.
William Rouse, Ph.D., Professor of Industrial & Systems Engineering at Georgia Tech, and Dietrich Stephan, Ph.D., the co-founder and Chief Science Officer at the personalized ge-nomics company Navigenics, addressed the workshop at the start of the second session. Rouse called healthcare a “complex adaptive system” and charted a systems-based approach focused on information management and the creation of incentives and challenges that would motivate stakeholders to provide quality, affordable care. Stephan placed his work at Navigen-ics, “the first fully integrated entity to make ‘personalized medicine’ a reality,” in the context of a future in which the latest knowledge is applied to prevent, delay onset, or cure disease. He de-scribed embedding one’s genome into the electronic medical record, with role-based access to the genomic data.
On the second morning of the workshop, Richard Bucholz, M.D., Professor and Director of Neurosurgery at the Saint Louis University School of Medicine, and Craig Feied, M.D., Chief Strategy Officer for Microsoft Research’s Health Solutions Group, provided “out-of-the-box” ideas to stimulate participants’ creativity. Bucholz presented his view of a future medical delivery system in which interoperable devices would unobtrusively, inexpensively, and efficiently collect, store, display, and exchange information through Web-based communication standards and protocols, facilitating improved process flows for patients, providers, and caregivers. Feied de-scribed “five forces” that would alter the healthcare landscape: truly definitive tests; systems biology approaches to unraveling the complexity of biological interactions; improved imaging techniques; evidence-based medicine; and true data liquidity.
Later in the day, Eric Horvitz, M.D./Ph.D., a Principal Researcher at Microsoft Research and a co-founder of Microsoft’s Healthcare Solutions group, gave a short presentation to lead off the final session. Horvitz demonstrated the promise of predictive modeling, including learn-ing models from the large amounts of data being generated, to inform decision-making in healthcare. He described a new advisory tool that helps inform clinicians’ decisions on whether to discharge a patient from the emergency room on the basis of the probability of that patient “bouncing back,” i.e., being readmitted to the hospital with a new diagnosis within a certain time window (days or weeks) in the future.
While the workshop covered a wide variety of topics, several themes ultimately emerged:
- Both technical and non-technical factors are causing healthcare to undergo major changes. Many aspects of healthcare are shifting to the home, and family members are taking on an in-creasing role as caregivers. Individuals are taking on a greater role in managing their own health. Professional caregivers are offered the promise of increased benefit from IT but the real-ity of inefficiencies and barriers make their jobs more difficult. The business of healthcare deliv-ery is becoming increasingly complex.
- The emerging technologies that can be used to improve health and healthcare are rich and diverse. Massive amounts of data about individuals, about the biology of disease and wellness, and about treatments and outcomes are becoming available in electronic form. Increasingly powerful techniques for data analysis are emerging. Sophisticated imaging techniques, sensing and monitoring technologies, and communication infrastructures are providing access to spe-cialized information in real-time. Robotic and speech technologies are enhancing human capa-bilities. Understanding of human behavior, awareness, incentives, and cognition is advancing.
- Both integration and specialization will play important roles. Health information, be it about an individual, a disease, or a therapy, and be it longitudinal or immediate, needs to combine data from multiple sources, multiple scales, and multiple representations. It must have high integrity, it must be comprehensive, it must be integrated over time, and it must be robust in the face of uncertainty and incompleteness. Access to that information must be contextual and use-driven. It must be easily and quickly understood. It must protect privacy, while not hiding what is needed for the situation.
The evolution of healthcare and the advances in information technology, broadly con-strued, create the need and also the opportunity for both long-term and short-term research. During the final session, the participants discussed some of the factors (apart from the obvious need for funding) that would facilitate research progress. The needs include the availability of publicly available de-identified data sets, open research infrastructures, for instance, for com-prehensive simulations, mechanisms for the migration of results to deployment, lowering of legal barriers to research, and appropriate forums to report the results of this multi-disciplinary research. A major challenge is to circumvent the cultural differences in research styles and research evaluation and to enable federal agencies with health IT-related interests to strengthen their ability to support meaningful collaboration and to lower the barriers for researchers to work together.
Susan Graham, Ph.D., (email@example.com) is Pehong Chen Distinguished Professor Emerita and Professor in the Graduate School in the Department of Electrical Engineering and Computer Science at the University of California, Berkeley, and was co-chair of the workshop organizing committee.
Erwin Gianchandani, Ph.D., is an American Association for the Advancement of Science (AAAS) Science & Technology Policy Fellow in the Directorate for Computer & Information Science & Engineering at the Na-tional Science Foundation, and helped with planning of the workshop.