DOE to use high-performance computing to analyze records of 24 million vets
The Department of Energy is leveraging high-performance computing and artificial intelligence to analyze the electronic health records—and other data—of more than 24 million veterans in an effort to improve their healthcare by developing new treatments and preventive strategies.
Suicide prevention, prostate cancer and cardiovascular disease are the three priority areas for this AI-driven Big Data Science Initiative (BDSI) designed to help accelerate medical breakthroughs.
Under BDSI’s Million Veterans Program-Computational Health Analytics for Medical Precision to Improve Outcomes Now, DOE and VA are creating a scientific computing environment that will house and provision data to researchers nationally, including access to the VA’s EHRs that span two decades.
“The VA has a unique dataset of medical records, whole genomes and imaging data that is one of the most comprehensive in dimensions of time, scale and breadth, and—in many aspects—this dataset is considered to be the largest and most comprehensive in the world,” testified Dimitri Kusnezov, chief scientist of DOE’s National Nuclear Security Administration, during a House hearing on Tuesday.
“This win-win enterprise could revolutionize quality of healthcare for veterans, while simultaneously providing DOE with unique insight and information to support development of next generation technologies,” said Rep. Barbara Comstock (R-Va.), chairwoman of the Research and Technology Subcommittee.
However, Kusnezov acknowledged in his testimony that the “challenge of the VA dataset” is that it “will uniquely stress our computers, codes and people in dimensions that our existing datasets have not.” On the policy side, he also noted that the Institutional Review Board (IRB) process—designed to protect the welfare, rights and privacy of human subjects in biomedical research—is something the agencies are “still working through” in order to allow access to the veterans’ data.
In addition, the initiative will tap into non-VA sources of data from the Department of Defense, Centers for Medicare and Medicaid Services and the Center for Disease Control’s National Death Index.
Nonetheless, according to Kusnezov, the United States is currently the only country with the ability to bring together a health database of this size with supercomputers and analytics capabilities that can “push the frontiers of computing and artificial intelligence.”
Rep. Lamar Smith (R-Texas), chairman of the House Science, Space and Technology Committee, said the DOE-VA partnership has the potential to transform the delivery of healthcare to veterans using complex computer models to learn more about the causes and warning signs of diseases.
“By providing DOE with access to a large-scale database, the VA will help the Energy Department develop next generation algorithms and modeling capability while ultimately providing the VA with data it can use to improve veterans’ quality of life,” added Smith.
At the same time, Kusnezov emphasized that both agencies are aware of the unique privacy and security sensitivities of the veterans’ health data.
“Our veterans’ data is unique and must be treated with the utmost concern for privacy and other national security considerations,” he concluded. “DOE provides a unique safe harbor for the data and applications of tools that is difficult to achieve in the commercial sector and with academic and commercial companies. Yet, creating safe and secure opportunities for these stakeholders to work with the data and apply the cutting-edge tools developed in their laboratories and companies is essential for success.”