The Department of Veterans Affairs (VA) has been using artificial intelligence and patient data as part of a suicide prevention program for veterans–a top clinical priority for VA. The REACH Vet program, started in 2017, uses predictive algorithms to identify risk factors for suicide in millions of veteran patient records for medications, treatment, traumatic events, overall health, and other information. It then uses the information to determine the top 0.1 percent of veterans at any facility at the highest risk for suicide in the next year. Clinicians then call these veterans for about an hour’s conversation, offering to help them create a mental health care plan.
In its first year (2007-8), the program reached more than 30,000 veterans and identified about 6,700 active VA users a month. According to the short article on findings published by the Suicide Prevention Resource Center in 2018, “veterans who engaged with REACH Vet were less likely to be admitted to an inpatient mental health unit, and more likely to attend mental health and primary care appointments compared to those not in the program. REACH Vet infrastructure includes a coordinator at every VA facility and a national team of clinicians who provide overall program support.”
There are pros and cons to this proactive approach–the pros being a reduction in veteran suicides and evidence of higher suicide risk in the three-to-six months of starting–and ending–an opioid prescription; and the cons being that some of the algorithms may be inaccurate–a veteran could be inaccurately ‘dinged’ for risk or a traumatic involuntary hospitalization. VA is still refining its algorithms in areas such as changes in medication dosage (including opioids) and clinical notes for mention of negative personal issues. POLITICO Health Care