ARISING HEALTHALIVE. AWAKE. AWARE.Read the magazines →

← November 18, 2026 edited reader edition

THE ARISING · Longevity & Healthspan

A federal clinical AI roadmap puts outcomes at the center of payment

HHS wants to test whether technology-supported care improves patients’ health. The roadmap does not require doctors to use AI or grant new authority to autonomous tools.

A federal roadmap lays out how the Department of Health and Human Services wants to evaluate, monitor and pay for clinical AI. It proposes starting with lower-risk uses, monitoring results as tools are used, and allowing more autonomy only when evidence supports it. It also calls for better data sharing with patient consent and clearer accountability when systems make mistakes.

One piece is already underway. In July, the Centers for Medicare & Medicaid Services launched ACCESS, a voluntary Original Medicare pilot for technology-supported chronic care. Participating care organizations are paid to manage qualifying conditions, and full payment depends on the share of their patients who reach defined outcomes. For high blood pressure, for example, the measure can be whether patients bring their blood pressure down, rather than how often they use an app. The model requires physician oversight and is meant to complement existing care.

HHS says it expects to extend that approach to more parts of primary care and wants to work with states on additional payment tests, including in Medicaid. Those are proposals, not a new nationwide rule paying for every AI-enabled service. ACCESS itself is not limited to AI.

On safety, a separate FDA discussion paper describes a potential way to assess certain generative AI-enabled medical devices before marketing. It pairs benchmark tests with clinical confirmation. It is not a requirement that all chatbots pass a physician’s exam. An ARPA-H project, ADVOCATE, is also seeking to develop and evaluate AI that could support cardiovascular care between visits. FDA authorization is a goal of that program, not something the project announcement says it has secured.

The practical question for a health system is whether a tool can produce better results than existing care without creating new problems. That means tracking what happens after an alert, who responds when the tool gets it wrong, and whether patients are healthier. A faster answer is useful; an improvement in care is what this payment model is meant to measure.

Read the federal clinical AI roadmap

Read the ACCESS model from CMS

Read the FDA discussion paper announcement

Read ARPA-H’s ADVOCATE program announcement

THE ARISING NEWSLETTER

Science moves forward.
Come with it.

Research, clear explanations, and a little room for wonder.
Read and follow The Arising on Substack.

Find The Arising on Substack ↗