Predictive Analytics and Veteran Suicide Prevention

One of the best things about working for the Department of Veterans Affairs was the vast amount of data available on Veterans receiving care through VA. While my research center often used surveys, focus groups, and interviews to collect data on Veterans, we frequently pulled in data from Veterans' medical records (with their permission, of course). And other researchers were accessing Veteran data directly to understand and improve care.

An issue we frequently heard about, and sometimes dealt with firsthand, was the high rate of suicide among our Veterans. Veterans are at high risk for many physical and mental conditions, and are at heightened risk for suicide. The National Suicide Prevention Lifeline was created to help anyone, including Veterans, who is feeling helpless. Through partnerships with VA and other federal agencies, we heard about many success stories of the Lifeline.

But with the vast amount of data available on our Veterans, it would be great if we could intervene and help before someone gets to the crisis point. Today, I read about how the REACH Vet program is using predictive analytics to identify Veterans at risk for suicide:

The REACH Vet program draws on the agency’s vast trove of electronic health records and uses predictive analytics to identify patients who might be at risk of suicide. It alerts VA clinicians of veterans who could benefit from more attention, and the program prompts clinicians to call and check in with their patients.

“What we found … not surprisingly is that veterans at highest risk of suicide are also at very high risk of some other things,” Aaron Eagan, VA’s deputy director for innovation, said Thursday at ACT-IAC’s Health Innovation Day in Washington. “They’re at significantly increased rates of all-cause mortality, accident morality, overdoses, violence … [and] opioids.”

Veterans who engaged with REACH Vet were admitted to mental health inpatient units less often, showed up to more mental health and primary care appointments and visited the VA more frequently, compared to veterans who weren’t part of the program.

Eagan said he expected veterans would be frustrated by the phone calls, but his team hasn’t gotten any complaints.

“It’s a great reminder that people really feel good about us caring about them, and that’s what the response generally is,” he said.

The REACH Vet team is updating its predictive model for the program now, and it’s starting a new collaboration with the Energy Department’s super computer, Eagan said.