Sponsors (2)
This bill funds AI-based predictive analytics to find and target veterans at suicide risk—potentially improving early detection and focused interventions—but raises significant privacy, transparency, coverage, and sustainability concerns because models may be proprietary, unequally distributed, and the pilot is time-limited.
Veterans at elevated suicide risk could be identified earlier because VA-funded AI models will analyze combined VHA clinical records and VBA benefits data to predict risk factors.
Local VA facilities and communities could receive targeted interventions and outreach when the pilot prioritizes high-risk areas, focusing resources where Crisis Line calls or suicide indicators are concentrated.
The program expands research and innovation capacity by funding at least two grants to nonprofit, academic, and private organizations to develop predictive analytics for veteran suicide prevention.
Veterans' sensitive personal and health data will be combined across VA systems for model training, increasing privacy, reidentification, and data-breach risks for veterans' PII/PHI.
Reliance on privately owned proprietary models with no mandated independent auditing limits VA control and transparency, making it harder to validate, audit for bias, or ensure safe clinical use.
The program's minimum-grant approach and prohibition on multiple grantees per VISN could leave many regions—especially rural or underserved areas—without locally tailored models or resources.
Based on analysis of 2 sections of legislative text.
Creates a VA grant program to fund entities to develop AI-based predictive models that assess veteran suicide risk and integrate VA benefits and clinical data.
Introduced April 23, 2026 by Ryan Mackenzie · Last progress April 23, 2026
Creates a temporary VA grant program, run by the Department of Veterans Affairs' Center for Innovation for Care and Payment, to fund eligible nonprofit, academic, private research, and other entities to develop AI-based predictive models that evaluate risk factors for veteran suicide. The VA must set up the program within one year, pick at least two grantees (subject to geographic limits), require integration of Veterans Benefits Administration data with Veterans Health Administration clinical data, allow grantees to retain proprietary model ownership, and end the authority on September 30, 2029.