While Congress has backed a controversial Department of Veterans Affairs program that has doubled down on artificial intelligence tools designed to identify veterans at high risk of suicide and has alerted more than 130,000 veterans since 2017, it has also drawn criticism for its treatment of women.
The Fiscal Year 2026 Military Construction and Veterans Affairs Funding Act signed into law last month includes language encouraging the VA to “use predictive modeling and analytics for veteran suicide prevention.” The measure allocates $698 million for suicide prevention programs as the department tackles a crisis that claims the lives of approximately 6,500 veterans each year.
At the heart of this push is REACH VET (Recovery Engagement and Coordination for Health). A machine learning system scans electronic medical records each month to identify veterans in the top 0.1% of suicide risk and alerts VA medical center coordinators to provide targeted support.
The issue of gender bias
In 2024, the program drew sharp criticism when research by the Fuller Project revealed that REACH VET's algorithm considered being a white male a stronger indicator of suicide risk than factors that primarily or exclusively affect women.
The system prioritized veterans who were “divorced and male'' or “widowed and male,'' but not groups of female veterans. Military sexual trauma and intimate partner violence, both associated with increased suicide risk among female veterans, were not included in the model at all.
The timing couldn't have been worse. Suicide rates among female veterans jumped 24% from 2020 to 2021, nearly four times the rate among male veterans, according to government data. The proportion of female veterans rose 10 times faster than women who had never served in the military.
“It's been tough being a woman in the military. We get harassed, we get bullied,” Air Force veteran Paulette Yazzie told reporters. “It is disgusting that the Veterans Administration can justify ignoring female veterans who are at risk of suicide.”
Reach VET 2.0
In response, the Department of Veterans Affairs launched REACH VET 2.0 earlier this year. As previously reported on NextGov, the updated model includes military sexual trauma and intimate partner violence as risk factors. They also removed race and ethnicity as variables to address concerns about racial bias.
Evan Carey, acting director of the Department of Veterans Affairs' National Artificial Intelligence Laboratory, told lawmakers at a hearing in September that the model “has been updated to ensure continued high performance in identifying veterans who are at the highest risk.”
This change is significant because one in three female veterans report experiencing sexual trauma in the military, compared to one in 50 men. Anonymous surveys suggest the real rate may be more than 2 in 3 women, suggesting significant underreporting.
A report from Disabled Veterans of America found that military sexual trauma, intimate partner violence, substance use disorders, and reproductive health issues all contribute to an increased risk of suicide among female veterans. The organization had asked the VA to revise its algorithm to take these factors into account.
humans still make phone calls
Charles Worthington, Virginia's chief technology officer and chief AI officer, told Congress that REACH VET “used AI algorithms to identify more than 130,000 high-risk veterans, improving outpatient care and reducing suicide attempts.”
However, the system does not work alone. Once a veteran is flagged, a dedicated coordinator at each VA facility reviews the alert on a central dashboard. We then work with healthcare providers to reach out to veterans, evaluate their care, and create an individualized safety plan.
Rep. Nikki Budzinski, ranking member of the House Veterans Affairs Technology Modernization Subcommittee, pressed Department of Veterans Affairs officials to promise that AI will never replace human intervention in suicide prevention. “As far as I know, there are currently no plans to use AI as a treatment device rather than a health care provider,” Carey responded.
Dr. Matthew Miller, Virginia Executive Director of Suicide Prevention, emphasized that REACH VET offers a “combination of innovation, technology, and humanity.”
The road ahead
REACH VET is just one of four AI initiatives focused on suicide prevention out of a total of 227 AI use cases created by VA in 2024. Other tools include a tool that analyzes clinical records for signs of suicidal thoughts and identifies veterans with both access to guns and active opioid use disorder.
The veteran suicide rate has remained stubbornly flat since 2008, when the Veterans Administration spent about $4.4 million on prevention efforts. The department currently spends nearly $700 million a year, yet about 17 veterans still die by suicide every day.
Whether AI can help bend that curve remains an open question. But with support from Congress and a revised algorithm that better takes into account women's experiences, REACH VET 2.0 becomes the Department of Veterans Affairs' most ambitious effort to date to leverage technology in the fight against veteran suicide.
The Veterans Crisis Line is available 24/7 for veterans in crisis. Call 988 and press 1, text 838255, or chat online at VeteransCrisisLine.net.
Stay on top of your veteran benefits
Military interests change all the time. Sign up for Military.com for everything from pay to medical, access to the latest pay charts and more, and all the latest benefits delivered straight to your inbox.
story continues
