Skip to main content
Premium Trial:

Request an Annual Quote

VA Collaborates With DOE on Supercomputing for Million Veteran Analytics


CHICAGO (GenomeWeb) – In a notable example of cross-department collaboration within the federal government, the US Department of Veterans Affairs is moving a key part of its Million Veteran Program initiative to the Department of Energy's supercomputing platform. The VA also will be counting on the DOE to crunch big data for suicide prevention and treatment of prostate cancer and cardiovascular disease.

"This is MVP and beyond," MVP Director Sumitra Muralidhar said of the nascent VA-DOE Big Data Science Initiative, which makes use of excess supercomputing and data science capacity in the Department of Energy's National Laboratory system.

"We are partnering with DOE to use their high-performance computing capabilities to allow thousands of researchers access to this unprecedented data resource over time in a secure environment. The transformative science that will be developed through this partnership will improve healthcare for veterans and all Americans," VA Secretary David Shulkin said in a statement in May.

The VA-DOE Big Data Science Initiative will pull in data from MVP and the VA's electronic health records system — containing records on 24 million vets going back more than 20 years — as well as health information from the Department of Defense, the Centers for Medicare and Medicaid Services, and the Centers for Disease Control and Prevention’s National Death Index. The supercomputing platform will apply predictive algorithms, artificial intelligence, and "deep learning," according to Muralidhar.

The two departments publicly announced the initiative last month, but collaboration really began last year as part of the Obama administration's Cancer Moonshot, according to Muralidhar. Some of that work has migrated to the private sector under the auspices of the Biden Foundation, though the bipartisan 21st Century Cures Act of 2016 provided ongoing federal funding for precision medicine as it applies to cancer research.

"Obviously, we need computing power," Muralidhar said. "But we don't need to reinvent the wheel within the VA," Muralidhar said.

Through the National Laboratory system, DOE has the infrastructure and the experience the VA is looking for. "They are experts at doing this type of thing with big data," Muralidhar said. This includes supporting genomic analysis in the Million Veteran Program Computational Health Analytics for Medical Precision to Improve Outcomes Now, or MVP-CHAMPION.

MVP, launched in 2011, aims to collect a variety of data — genetic, medical, lifestyle, and military exposure information — from 1 million vets in order to understand risk factors related to post-traumatic stress disorder, schizophrenia, bipolar disorder, and other widespread conditions in that population. The goals of the MVP are similar to the Precision Medicine Initiative, and MVP has agreed to help recruit veterans interested in joining the national effort to advance personalized medicine research.

After President Barack Obama launched PMI two years ago, the White House gave the department a goal to recruit half the MVP cohort by the end of 2016. The VA met that goal and now has more than 567,000 veterans enrolled, according to Muralidhar.

The project only has enough funding to sequence whole genomes of 30,000 subjects in a two-year period that started in January. About 2,000 had been completed as of last month, though the VA is pushing forward with exome sequencing when WGS is not feasible. "We have done exome sequencing on about 30,000 samples," Muralidhar said.

"As our budget allows, we're going to be getting more and more in-depth data on these samples, cleaning them to the extent possible, and making them available to researchers," she had previously said.

Muralidhar noted that DOE has a "PHI enclave" in its supercomputing platform, secure enough to safeguard protected health information to HIPAA standards. Researchers will only see deidentified, coded data anyway, she said.

Early pilots of the VA-DOE Big Data Science Initiative will target patient-specific analysis for suicide prevention, prostate cancer, and cardiovascular disease.

2016 VA study found that about 20 veterans commit suicide daily. Researchers involved in the DOE collaboration will work with the VA Office of Suicide Prevention to refine algorithms that generate risk scores to help clinicians identify the most troubled vets, department officials said.

For prostate cancer, the VA hopes to cut down on unnecessary surgeries that affect cost, patient safety, and quality of life. "Right now, doctors can't tell the difference between lethal and nonlethal forms of the cancer without doing surgery, so they do the surgery anyway," Muralidhar said.

Cardiovascular disease is one of the most expensive components of VA healthcare. "No one has looked at everything together to look at who will develop what," Muralidhar said. She hopes the genomic and big-data analysis will improve understanding of risk factors and lead to better methods for determining individual risks and therapies.

As for the DOE, "They need a win, too," Muralidhar said. "It's going to push their development of exoscale computing."

This, according to Muralidhar, should result in better healthcare, better science, and better government.

The hope is to launch the pilots by late fall. "The environment is being built right now at DOE," Muralidhar said.

The DOE did not respond to multiple requests for comment on this collaborative project.