AI could revolutionize health care, but regulation is needed, FDA commissioner says

© Have a nice day - stock.adobe.com

Artificial intelligence (AI) could revolutionize medicine – so regulation is crucial when more computers have more interaction with physicians and patients, said the head of the U.S. Food and Drug Administration (FDA).

Robert M. Califf, MD, MACC, commissioner of food and drugs for FDA, spoke this month as part of the National Health Council’s 2023 Science for Patient Engagement Symposium in Washington, D.C. The conference, titled “Patient Empowerment in the Digital Health Era,” dealt with “an exciting set of ‘red hot’ topics,” including AI, he said.

Robert M. Califf, MD, MACC
_{U.S. Food and Drug Administration}

FDA and patients

FDA has gained experience integrating patient voices into the administration’s work through projects such as the Patient-Focused Drug Development initiative and the Patient Representative Program. Those specifically add patient and caregiver perspectives, including inviting patients to serve on FDA advisory committees and to present at FDA meetings and workshops, Califf said.

Joining with AI

As for integrating technology, researchers “have enormous potential,” entering a fourth Industrial Revolution with rich and diverse sources of digital data available in real time, Califf said.

Those include the “large language models” of AI, which generate words and texts based on huge databases and parameters.

“Large language models are the next step that appears to be ushering in the revolution that many of us were hoping for,” he said.

The ultimate goal of gathering data and evidence is to inform physicians’ and patients’ knowledge about the benefits and risks of medical interventions, to improve health relative to risk of harm, Califf said.

“If we’re not nimble in the use and regulation of large language models, we’ll be swept up quickly by something that we hardly understand,” Califf said. “The great things are really great: Imagine a world in which your questions were answered immediately in language appropriate for your literacy and numeracy; also your clinician can actually talk with you rather than spending all their time cutting, pasting and writing clinic notes; I could go on and on, but I see the regulation of large language models as critical to our future.”

Patients, tech, and society

More data will drive development of medical products and treatments. It has important implications for food safety, tobacco products, and all the medicines that FDA regulates, Califf said.

And there are other issues at the intersection of medicine and technology:

• Disparities. U.S. life expectancy is declining partly due to COVID-19, but also because of gun violence, suicide, opioid overdose, with stark differences based on race, ethnicity, sex, education, wealth, and rural living status.
• Representativeness. Ubiquity of cell phones show technology is “embedded in our human interactions,” Califf said. Now researchers must measure diversity of values, preferences, believes, living circumstances, and economic factors.
• Algorithmic bias. Once put into use, algorithmic computer programs will evolve and will need to be updated and adjusted.
• Misinformation. People connect through social media networks, but AI gives “almost everyone” the potential to produce false narratives with fabricated images and voices.

As private industry finds new ways to insert machine learning and mathematical algorithms into everyday life, government regulations lag behind, Califf said.

“Quite simply, we need to assemble the resources to put in place these policies and tools and adaptively align our digital health efforts to support public health and regulatory innovation in a world that is changing rapidly,” he said. “The potential certainly exists, and we have good reason to hope that we will fulfill the enormous potential of digital health technologies and see more innovative solutions that can result in improved outcomes, but we’d better also preempt the risks.”