A federal health research agency is committing $62.7 million to develop artificial intelligence systems designed to help manage heart failure, where heart failure means the heart muscle can no longer pump enough blood to meet the body's demands. The program, called ADVOCATE, aims to produce AI tools the Food and Drug Administration would authorize to assess symptom severity, prescribe drugs, and order lab tests. ARPA-H, the Advanced Research Projects Agency for Health, announced the first round of awards on Wednesday.
What authorized AI in cardiology would actually mean
Most AI tools in medicine today function as advisories. A physician receives a recommendation, reviews it, and decides. ADVOCATE is aiming at a different level of machine involvement: partially autonomous devices, meaning software capable of initiating clinical steps within regulatory limits, without requiring a human decision at each point.
Getting there requires FDA authorization for that level of autonomy, which is among the central goals the funded teams must clear.
The first year of the program carries $33.7 million in committed awards. The remaining portion of the $62.7 million total is not yet locked in. ARPA-H has said the balance may be renegotiated up or down.
Who received the first round of funding
The initial awards went to health technology companies Atman Health, UpDoc, and Tempus AI. Research teams from Stanford University, Duke University, and the Kaiser Permanente health system also received funding in this first group. ARPA-H has indicated it may still fund additional teams beyond this initial batch.
The gap the program is built around
Heart failure affects 6.7 million Americans. ARPA-H's stated rationale for ADVOCATE is that many of those patients do not receive optimal treatment, and the primary barrier is specialist access.
Cardiologists and advanced heart failure programs tend to concentrate in urban and well-resourced medical centers. For patients in rural communities or areas with limited specialty care, that distance translates into less consistent management.
ARPA-H said the hope is that AI agents developed under ADVOCATE could operate in those settings, assessing symptoms, prescribing drugs, and ordering lab tests in places where a specialist visit is not practical. Whether the funded teams can produce devices that meet the FDA's standard for that level of independence is what the program's remaining timeline will determine.