AHCJ – Association of Health Care Journalists

09/11/2024 | News release | Distributed by Public on 09/11/2024 13:46

What to know about FDA’s new Digital Health Advisory Committee

Photo by Karolina Kaboompics via Pexels

The U.S. Food and Drug Administration recently created its first advisory committee for digital health. The emergence of digital health over the last several years and the recognition that some of these tools and applications inform direct health care delivery spurred this decision, said committee member Thomas Maddox, M.D. Dr. Maddox was a panelist at Health Journalism 2023, where he discussed how AI could change health care delivery.

"Most of the impetus in forming this committee now is the recent explosion of AI in health care," Maddox said. "If we thought digital tools were hard to regulate, AI is going to add another level of complexity. Part of our charge will be to think through this technology and where it's headed, how we ensure its safe and effective use in health care, and what the FDA's role is in that."

The new group will advise the FDA about complex scientific and technical issues related to digital health technologies such as artificial intelligence/machine learning (AI/ML), augmented reality, virtual reality, digital therapeutics, wearables, remote patient monitoring and software. The committee will also provide expertise and perspective to improve the FDA's understanding of the benefits, risks and clinical outcomes associated with using these technologies, according to an FDA website.

Journalists interested in digital health technologies could follow along with the group's work, including its first meeting planned for November, and cultivate sources from its experts.

More about the committee

The committee comprises nine voting members, consisting of academic experts, with input from a roster of leaders from the private sector, including executives from Medtronic, Boston Scientific and Johnson and Johnson. Industry representatives without voting privileges have expertise in various subtopics like hospital system integration of digital health technologies, digital diagnostics, wearable devices, personalized medicine and more.

The agency's first committee meeting will be November 20-21 in Gaithersburg, Md. Members will discuss how generative artificial intelligence (AI) may impact the safety and effectiveness of medical devices enabled with this technology. Generative AI is AI that can create original content in response to user prompts. Further details about the meeting, which will be webcast, will be provided in the Federal Register in the coming months and likely will be available on the committee's website.

The FDA has been clearing AI devices "at a blistering pace," STAT's Health Tech e-newsletter editors wrote on May 16. Some 150 such devices were approved between August 2023 and March 2024.

A lot of generative AI activity is occurring in clinical documentation, Maddox said. For example, ambient listening programs are being trialed at his and other medical centers that record doctor-patient conversations during outpatient visits and feed a transcript of the conversation into a large language model, which in turn transforms it into a summarized note of the clinical encounter for a provider to review, correct if necessary, and sign off.

"It's been pretty impressive - the accuracy of the AI-generated clinical notes is quite high," Maddox said. At his institution, providers were accepting about 85% of draft notes and sharing anecdotes about how the tool freed up time.Now, they can eat dinner with their families more often.

AI in clinical settings

AI technology could soon extend to other clinical settings, he said, such as nurse interactions with hospitalized patients or doctor-patient-family conversations in the emergency department. "You can imagine all the different venues where that will be useful," he said.

Down the road, clinicians could use AI to gather ideas about the causes of patient symptoms or the most effective management strategy for a given clinical condition, Maddox said, putting it in the sphere of clinical decision support. "Like any tool that we use for clinical decision support, it needs to be accurate, it needs to be used appropriately, and it needs to make sure there's a human in the loop to ensure safety."

Trends to follow

Maddox advised journalists to watch for the following trends:

  • How can AI help with clerical tasks that are causing burnout among health care clinicians?
  • Generative AI is being used in some cases to help draft letters and responses to patient emails. For now, Maddox said, this is largely for routine questions like asking for patient refills of medications. But how could the technology be incorporated into more medical management-type conversations with patients?
  • How could AI programs be used to comb through patient health records to identify opportunities for treatments or screening, as a double-check against clinicians' work?

Resources