Proposal

The National Research Council’s Precision Medicine report ^[1] found that it is imperative to create a new scientific base for biomedical research, clinical care, and public health that accurately reflects the genetic variations in diseases and in individual responses to therapies.

This proposal calls for using the nation’s rapidly expanding capabilities for computerized biomedical research to accomplish this goal as quickly as possible. Research databases and analyses for most diseases would be completed over the next three years (by the end of 2015).

Background

The US-led Human Genome Project was finished ten years ago (2003).  In the past several years, key elements for moving forward on a Precision Medicine-type initiative have been coming together.
.

• Investments in large biobanks have linked individual genetic data to electronic health records. Examples include the Kaiser-Permanente biobank with 500,000 individuals (supported by the National Institutes of Health), the Department of Veterans Affairs’ Million Veterans biobank ^[2], the Human Genome Institute’s ENCODE network, and the UK’s national biobank (500,000 patients).
• Research advances have shown that much of today’s disease taxonomy is wrong, i.e. it is symptom-based and groups together diseases that have different genetic and other causes. It leads to giving the same treatments to individuals who actually have different diseases and respond differently to treatments. Thus a new era of precision diagnoses will be essential for a new era of targeted, more effective therapeutics. Cancer ^[3] seems to be one of the most promising areas for such “precision medicine” initiatives.
• A new policy consensus is developing for a “rapid learning health system ^[4]” that uses the power of high-speed computers (today’s peta-flop computers perform a quadrillion operations per second) applied to electronic health record databases with millions of patients, biobanks and other registries to advance comparative effectiveness research and personalized health care.
• An understanding has developed that America, for its health and economic future, needs to invest in being the world leader in biomedicine and in using computers to translate big data quickly into useful knowledge.

Specifics

With such start-up capabilities in place, the US can now launch a patient-centered initiative for personalized medical care to benefit most of our population. As envisioned in the Precision Medicine report, the short-term and future results of this new knowledge network would revolutionize biomedical research, clinical care and public health.

Here’s one way this might be done.

The Department of Health and Human Services (NIH) could lead the initiative. This BD2K (Big Data To Knowledge) strategy would capitalize on national research laboratory models (the Department of Energy’s physics research system) and government-directed research initiatives (like the Department of Defense-Defense Advanced Research Projects Agency (DARPA) and the Human Genome Research Project), rather than relying on the traditional NIH-RO1 investigator-initiated mechanisms.
.

• Advisory panels would be convened for advancing precision medicine in major disease areas, e.g. perhaps 20 individuals for each panel from researchers, clinicians, patients, biotechnology companies, health plans and delivery systems, computational sciences, government agencies, and others.  There should be one or more advisory panels focused on patients with multiple conditions; these patients are often excluded from clinical trials but account for a large share of medical care spending. An overall coordination group might be desirable, as there will be common methodological, computational, scientific, and reporting issues for many of the projects.
• Project directors and research teams/centers would be recruited to lead and conduct the research, on behalf of the advisory group. They would obtain and analyze the best available data on genetic variations in diseases and in treatment responses from US and other biobanks, EHR databases, and other resources. They would learn as much as possible, as quickly as possible, from data mining rather than conducting new clinical trials. Initial results would be considered; several rounds of refining studies might be required. Advisory committees could use “crowd-sourcing” invitations and national challenges to obtain fresh looks at these data. NIH would be responsible for assuring availability of data for the projects, scientific and clinical usefulness of the studies, and overall success of the projects.
• By the end of the three-year project (the end of 2015), each project director/team/advisory committee would publicly report their work, their findings on genetic variations in diseases and treatment responses, their suggestions for a new disease taxonomy, and thoughts about the next steps in the precision medicine research program.  Their new (de-identified) databases would reside at NIH’s National Library of Medicine, where they would build a world evidence base for open science and future “precision medicine” research.

While existing databases have good coverage for adult medicine, it might be necessary to quickly augment biobank resources for pediatrics, particularly for special-needs children often found in the Medicaid program. NIH might also need to give special attention to collecting more EHR registry and genetics data for relatively rare diseases, for minorities, and for co-occurring conditions in high-needs populations.

Conclusion

A “precision medicine” initiative is imperative to advance biomedical research, clinical care, and public health. This proposal would use leadership in the Human Genome Project and new national capacities for rapid learning to make dramatic progress over the next three years. The resulting knowledge and infrastructure would expand America’s leadership in biomedical and computing sciences.