Inside HHS's AI push: Department efforts align around governance, collaboration

ARPA-H tackles cardiovascular disease with agentic AI

Haider Warraich, M.D., a cardiologist and program manager at the Advanced Research Projects Agency for Health, highlighted ARPA-H's Agentic AI-EnableD CardioVascular CAre TransfOrmation, or ADVOCATE, program, as an example of how HHS is prioritizing AI adoption in clinical care.

The ADVOCATE program was born out of the need to address preventable deaths from cardiovascular disease, the leading cause of death in the U.S., Warraich said.

"And one of the biggest reasons is that there simply is not enough cardiology care available. Research that my team led in the past showed that almost half of the counties in this country don't have a single cardiologist," he noted.

"And even when people do have access to cardiologists, the type of care or the quality of care that they receive is often not what you would want, or not what I would want for my patients or my loved ones. And so certainly there is an opportunity for AI to be able to scale clinical knowledge into action broadly."

Clinical AI agents could help fill this gap and improve cardiovascular disease management, Warraich asserted, with 24/7 assistance on tasks like appointment changes and diet and exercise guidance.

"The vision for ADVOCATE was really to support teams that will develop a technology that can essentially do everything a clinician can do over the phone," Warraich noted.

Beyond administrative tasks, ARPA-H envisions the AI agents eventually performing high-risk functions, including coordinating triage, prescribing new medications and providing diagnoses.

ARPA-H demonstrated industry-wide collaboration by opening the ADVOCATE program up to leaders in technology, academia and nonprofit organizations. The program solicited these experts to submit proposals for the development of a patient-facing clinical AI agent, a supervisory agent that ensures the clinical agents' safety and efficacy and a scalable plan for integrating these tools into clinical workflows.

The overarching goal of the ambitious program is to create an FDA-approved clinical agentic AI system that serves as an extension of the care team. ARPA-H is actively reviewing proposals now.

Beyond administrative tasks, ARPA-H envisions the AI agents eventually performing high-risk functions, including coordinating triage, prescribing new medications and providing diagnoses.

"I think many of us see the potential of AI, and the more you study it, the more you see new advances, the more you realize that this is a technology that could well be the defining technology of our age," Warraich said.

However, he reasoned that patients have not yet seen AI impact their care outcomes, citing regulatory uncertainty and the high risk of developing AI technologies as primary reasons.

Warraich expressed optimism about the future of AI in clinical care if the ADVOCATE program succeeds.

"The focus of this program is on patients with heart failure, but really the idea is that we are building a template that would be applicable to any patient with chronic disease," he added. "And so, the impact could be quite broad."

Administration for Community Living helps individuals age in place

The Administration for Community Living, a small HHS department that funds community-based services and research to empower older adults to live independently, demonstrated how it is using AI to keep people comfortable and safe in their homes.

Mary Lazare, the principal deputy administrator for the ACL, said the department is looking to use AI to address a range of issues affecting its target communities.

"We're beginning to look at that in terms of wearables, but also modifying homes so that people can remain in their homes -- safety features, ease features, locking doors, turning on lights, all of that -- and scheduling that through AI so that it becomes manageable certainly for people with disabilities and those who want to continue living in their homes into older age," Lazare said.

The ACL is also looking to use AI to help people with disabilities get jobs by connecting them to employers seeking specific skill sets, detecting scam calls and preventing abuse.

Kelly Cronin, deputy administrator for innovation and partnership at the ACL, highlighted two programs that the ACL has stood up to accelerate AI. The first is a caregiver AI prize challenge. Currently in phase one, the challenge is seeking practical ways to use AI to fill the widening caregiver gap.

Rewards will be given to those who can create AI tools for caregivers and to extend the caregiver workforce, reducing burden and improving care quality in the process.

"Importantly, we want to supplement, not replace, human connection," Cronin said. "And we also want AI tools to be user-centered, really informed by the voices and experiences of caregivers."

The second AI initiative is the ACL's multi-phase Health at Home Challenge, which focuses on supporting dually eligible Medicare and Medicaid beneficiaries at home through comprehensive services. The challenge participants must demonstrate how advanced community care networks and healthcare partners can improve affordability and independence at home for these dually eligible beneficiaries.

"We recognize technology infrastructure and tools are really going to be an important enabler of this work," Cronin said. "And in designing this and now moving into these next phases, we want to reward the use of AI, particularly AI that's going to enable that clinical and community integration and support care coordination across the continuum."

The Health at Home challenge will enter phase two in August 2026, wherein up to $2 million will be awarded to a maximum of five winning teams from phase one to accelerate tool implementation.

FDA focuses on policy development

The FDA also provided its stance on AI during the webinar, albeit with less specificity than the other departments.

Rick Abramson, M.D., the recently appointed director of the FDA's Digital Health Center of Excellence, said he would share the FDA's agenda at a high level "because we are engaged in active policy development, which does somewhat limit my ability to speak to specific policy directions."

After hinting at upcoming policy shifts, Abramson detailed the FDA's approach to regulating clinical AI. After conducting public workshops and listening sessions, Abramson said the FDA is focused on responding to stakeholders' calls for greater clarity.

"This has always been important and it will be even more important as we head towards AI systems of greater functional complexity and AI systems that operate with greater autonomy and greater agency," he noted.

"FDA will need to be clear not only on what it regulates and how, but also on what role it will play in both the pre-market and post-market settings. And we aim to get that clarity out to the community."

Abramson also stressed the importance of right-sizing regulation in proportion to risk and providing appropriate oversight across the lifecycle of a product. He further noted the need for policy coordination between federal entities, state oversight bodies and international regulators.

"The public should be expecting us to release some ideas to the public for stakeholder comment in very short order," Abramson said, showing that, like other HHS departments, the FDA is aiming to accelerate clinical AI adoption while prioritizing risk management.

Jill Hughes has covered health tech news since 2021. Her coverage areas include cybersecurity, HIPAA compliance, interoperability, AI and EHRs.