DistilINFO
Recent Posts
HomeGovHealthHHS Backs Autonomous AI Agents for Clinical Care

HHS Backs Autonomous AI Agents for Clinical Care

Why HHS Is Pushing for Autonomous Clinical AI

For years, AI played a supporting role in clinical care. It flagged suspicious scans. It drafted notes. It summarized charts. Today, however, federal health officials are reaching for something far more ambitious: AI agents that act on their own — diagnosing patients, adjusting medications, and closing the loop without a physician in the middle.

This vision surfaced publicly on June 25, when senior leaders from across the Department of Health and Human Services gathered online to share findings from a December request for information on accelerating AI adoption in clinical care. That RFI attracted more than 7,000 public comments — a signal that the healthcare industry is watching closely. Notably, HHS organized the event around a coordinated, cross-agency approach it calls “one HHS,” building on an AI strategy the department released late last year.

Thomas Keane, MD, national coordinator for health IT, opened the session by recounting how AI now helps physicians detect strokes on imaging in seconds — work that was effectively impossible two decades ago. “Technology has really transformed clinical care, and a big part of that technology is AI,” he said. Yet the tools Dr. Keane described still assist a physician. They do not replace one.

That distinction is exactly what the next phase of this effort intends to challenge.

What the ADVOCATE Program Actually Does

ARPA-H’s Bold Bet on Agentic AI

The most concrete vision for fully autonomous clinical AI came from the Advanced Research Projects Agency for Health (ARPA-H). Haider Warraich, MD, a cardiologist and ARPA-H program manager, is overseeing a program called ADVOCATE — short for Agentic AI-Enabled Cardiovascular Care Transformation.

The program’s goal is to build the first FDA-authorized agentic AI capable of delivering around-the-clock cardiovascular care. This is not a chatbot. Furthermore, it is not a decision-support tool that routes every recommendation back to a clinician. Instead, ADVOCATE aims to build software that can act independently, in real time, on behalf of the patient.

A Template for All Chronic Disease

Although ADVOCATE focuses specifically on patients with heart failure and those recovering from heart attacks, Dr. Warraich made clear the ambitions extend further. “We are building a template that would be applicable to any patient with chronic disease,” he said.

He described the overarching goal as building “the entire stack that will be needed to make clinical agentic AI a reality for every American.” That framing positions ADVOCATE not as a niche cardiology tool but as a proof-of-concept for the entire U.S. healthcare system.

How the Technology Would Work at the Bedside

Real-Time Integration with EHRs and Wearables

The ADVOCATE AI would be patient-facing and connected in real time to the electronic health record. It would also pull data continuously from wearable devices. This live data stream allows the agent to monitor patients around the clock — far beyond what any clinician could realistically provide.

The AI’s clinical scope would go well beyond simple scheduling or lifestyle coaching. Specifically, Dr. Warraich outlined capabilities that include:

  • Diagnosing patients based on incoming data
  • Performing triage and determining urgency of intervention
  • Prescribing or adjusting medications — including higher-risk decisions — without routing back to a clinician

“This technology will be able to close the loop, instead of directing them to a clinician,” Dr. Warraich said. “It will be able to take the actions that the patient needs.”

A Second Agent to Watch the First

Recognizing that even autonomous AI can make errors, ADVOCATE also funds a second “supervisory” agent. This oversight layer monitors clinical AI agents after deployment and flags situations where human intervention is needed. As Dr. Warraich acknowledged, “Clinicians will have a lot of concern about technology that is so powerful.” Therefore, building in a mechanism for human oversight is central to making the system trustworthy.

The Access Gap Driving This Initiative

Nearly Half of U.S. Counties Lack a Cardiologist

The core rationale for autonomous clinical AI is not efficiency. It is access. Dr. Warraich pointed to a stark statistic: nearly half of all U.S. counties do not have a single cardiologist. In those communities, patients with heart failure or post-heart-attack conditions may wait weeks — or simply go without specialist care entirely.

The goal, accordingly, is software that can do “everything a clinician can over the phone.” For rural and underserved communities, that capability represents a meaningful leap in care access. Moreover, it shifts the conversation from AI as a productivity tool to AI as a care-delivery mechanism in its own right.

FDA Is Building a New Regulatory Framework

Proportionate Risk Regulation

Autonomous clinical AI creates regulatory challenges that current frameworks were not designed to handle. Rick Abramson, MD, director of the FDA’s Digital Health Center of Excellence, addressed this directly at the HHS event. The agency is actively preparing for AI tools that operate with “greater autonomy and greater agency,” he said.

The FDA’s approach centers on what Dr. Abramson called “regulation that is proportionate to risk.” Crucially, he stressed that oversight cannot stop at the point of approval. Clinical AI tools, he noted, can “behave very differently in the post-market setting from how they behaved pre-market.” Consequently, monitoring after deployment is as important as the approval process itself.

Dr. Abramson indicated the FDA plans to “release some ideas to the public for stakeholder comment in very short order,” though he declined to detail specific policy directions while they remain under development.

What This Means for Health Systems Today

Autonomous AI Is Coming — The Rules Are Still Being Written

For health system leaders, the message from HHS is clear: autonomous clinical AI is no longer a distant concept. It is an active federal priority, backed by research funding and a coordinating regulatory strategy across multiple agencies.

Still, several hard questions remain unresolved. Chief among them is accountability. When an autonomous agent makes a clinical error, who is responsible — the hospital, the vendor, the federal agency that authorized it? These are not hypothetical scenarios. Some health systems are already grappling with liability questions as they deploy earlier generations of clinical AI.

Timeline and Clinical Availability

ADVOCATE is currently a research and development program. It is not a product available to patients today. ARPA-H has outlined a multiyear timeline that includes early rounds of evaluation before any clinical trials begin. In short, health systems have time to prepare — but the regulatory and operational groundwork must start now.

The unresolved challenge, which the agencies themselves acknowledged they are still working through together, is how to let an AI act on a patient’s care without losing the trust of the clinicians it is meant to extend.

Written by

The DistilINFO Editorial Team covers healthcare payer, provider, government health, life sciences and health AI news for more than 135,000 industry decision makers. Since 2016, DistilINFO has published daily briefings that summarize the developments that matter most to health plans, health systems and the organizations that serve them. Our stories are based on primary sources, including press releases, regulatory filings, CMS and HHS announcements, and peer-reviewed research, which we cite and link in every article. Questions or corrections: digital@distilinfo.com.