The Advanced Research Projects Agency for Health (ARPA-H) announced a commitment of about $62.7 million to a program named ADVOCATE aimed at developing artificial-intelligence agents to assist in the treatment of heart failure. ARPA-H described the program as targeting the creation of partially autonomous AI devices that would be eligible for authorization by the Food and Drug Administration (FDA). The agency awarded initial funding to a first set of teams and indicated it may fund additional groups in the future.
The announcement specified that the amount committed for the program’s first year is $33.7 million, with the remainder of the $62.7 million subject to renegotiation upward or downward. The statement did not provide details about the total planned budget distribution across years, milestones tied to payments, or criteria for future awards beyond the first tranche.
ARPA-H described the ADVOCATE program’s goals as producing software-based agents capable of clinical functions that today are typically performed by clinicians. The tasks explicitly mentioned in the announcement include: assessing the severity of patients’ symptoms, prescribing medications, and ordering laboratory tests. ARPA-H used language indicating the ambition to create “partially autonomous AI devices” and referred in coverage to the concept of an “autonomous prescribing cardiologist” as an aspirational role for these systems.
The announcement framed these AI agents as clinical tools that could help manage heart failure patients by performing assessment and treatment actions. The source did not provide granular detail about how autonomy would be defined, which prescribing authorities would be delegated to the agents, or specific clinical decision-making algorithms. It also did not specify clinical endpoints, safety thresholds, or monitoring strategies that will be required during development and prior to any FDA authorization.
ARPA-H announced the first set of awards to a mix of private companies and academic or integrated health system teams. Named recipients in the first round include the health technology companies Atman Health, UpDoc, and Tempus AI, as well as research teams from Stanford University, Duke University, and the Kaiser Permanente health system. The agency said it may fund additional teams beyond these initial awardees.
The announcement did not disclose the specific award amounts assigned to each team, the particular roles or technology approaches each recipient will pursue, or whether the awards fund clinical trials, engineering development, human factors research, regulatory work, or implementation studies.
The only financial specifics supplied in the announcement were the total program commitment of approximately $62.7 million and the first-year allocation of $33.7 million. ARPA-H noted that the remainder of the committed funds could be renegotiated. The source did not include a project timeline for milestones, expected duration of the ADVOCATE program, or dates for when prototype systems, pilot studies, or FDA submissions might occur.
There was no discussion in the announcement of post-authorization support, commercialization plans, reimbursement strategies, or how ARPA-H will evaluate clinical benefit and safety to justify continued funding.
ARPA-H positioned the ADVOCATE program as an intervention to help close gaps in care for the large population of people living with heart failure. The announcement cited that roughly 6.7 million Americans have heart failure and emphasized that many of these patients do not receive optimal treatment because of limited access to specialists. ARPA-H portrayed AI agents as a means to expand the delivery of guideline-based assessment and therapeutic actions into rural and other underserved settings where cardiology access is limited.
The source did not provide data within the announcement quantifying how AI-driven interventions are expected to change outcomes, or modelling of anticipated impacts on access, adherence, hospitalizations, or mortality.
The ARPA-H announcement conveyed high-level goals and named initial awardees, but many operational, clinical, and regulatory details were not reported in the source article. Absent specifics include:
Because those details were not reported in the source article, readers should consult ARPA-H’s program documents or direct communications from awardees for more complete technical, regulatory, and operational information.
Summary
ARPA-H’s ADVOCATE program represents a targeted, multi-million-dollar effort to advance FDA-authorizable, partially autonomous AI agents for heart failure care. Initial awards went to a combination of health tech companies and academic/health system teams; first-year funding was specified but many practical and regulatory details about development, evaluation, and deployment were not included in the announcement and remain to be disclosed by ARPA-H or the funded teams.