At HLTH's health care industry conference in Las Vegas, HHS officials said they’re developing a cross-agency strategic plan to regulate artificial intelligence in health care, aiming to balance its benefits and potential risks.
Key to that effort will be hiring a chief technology officer, a chief data officer and a chief AI officer. “We’re entering an era where we have to rethink everything we do,” Dr. Robert Califf, commissioner of the HHS’ Food and Drug Administration, said at the conference last week. The backstory:
The Department of Health and Human Services and its agencies regulate AI when it’s embedded in medical devices and when it poses a risk to privacy or the potential for discrimination. In December, the agency’s assistant secretary for technology policy finalized rules requiring AI developers who want federal certification to reveal additional data about their algorithms. The FDA is also starting to use AI internally for compliance monitoring and thinking about how hospital systems can regularly validate AI, Califf said.
But so far, HHS and its agencies have taken a flexible approach to AI regulation, issuing more guidance than rules and working with the health tech industry to build safe products. Officials restated that approach in Las Vegas. “A lot of my cases you don’t read about in the news, you don’t read about them in POLITICO, because they’re handled with technical assistance, voluntary compliance,” said Melanie Fontes Rainer, acting director of HHS’ Office for Civil Rights.
Why it matters: Industry wants guidance on how regulators will handle artificial intelligence so they can build compliance systems early. Several industry groups, most prominently the Coalition for Health AI, which counts Microsoft, the Mayo Clinic, Johns Hopkins and Duke among its members, have formed to pursue private sector-led vetting of AI tools.
What’s next? HHS officials have said private sector-led compliance labs can be part of the answer. Califf has repeatedly said the FDA doesn’t have the resources or personnel to vet all advanced AI tools. In Las Vegas, he said the health care industry doesn’t, either. “I don’t know of a single health system in the U.S. which is capable of doing the validation,” he said. “We need a major change.”
Fontes Rainer said her agency is considering asking health systems and payers to submit to voluntary audits to ensure compliance with her anti-discrimination rule. “We’re thinking about doing that because AI is so different when it comes to regulation and it comes to enforcement,” said Fontes Rainer. She said her office is working on additional guidance and hopes to share more of its thinking on AI before the end of the year.
Even so: Everything could change with a new administration. Former President Donald Trump has promised that if he wins he would revoke the executive order President Joe Biden issued a year ago, tasking federal agencies with launching AI rules.
| 
