Artificial intelligence (AI) could be used to prescribe medications to patients — if a new bill makes its way through Congress.
The proposed legislationopens in a new tab or window, sponsored by Rep. David Schweikert (R-Ariz.), would amend the Federal Food, Drug, and Cosmetic Act to clarify that AI and machine learning technologies can qualify as a practitioner eligible to prescribe drugs if authorized by the state involved and approved by the FDA.
The bill was introduced in the U.S. House of Representatives and referred to the House Committee on Energy and Commerce in January.
But Adam Rodman, MD, MPH, a hospitalist and director of AI programs at Beth Israel Deaconess Medical Center and assistant professor at Harvard Medical School in Boston, told MedPage Today that the technology is not nearly where it needs to be for this kind of prescribing.
However, “[t]hings are accelerating so quickly,” he added, “I don’t doubt that we will be having this conversation,” at some point.
He said he believes the proposed legislation “reflects enthusiasm about the technology,” and that it could be aimed at helping to address “huge care gaps,” when people struggle to see their doctor.
Still, “there would have to be huge sets of regulations that govern this,” Stephan Fihn, MD, MPH, an internal medicine physician and professor in the departments of medicine and health systems and population health at the University of Washington, told MedPage Today. As it is, the proposed legislation doesn’t specify which drugs it could be used for and in which settings, or whether physician review would be required.
Currently, there are many applications for AI in medicine, and “they’re increasing very rapidly,” said Fihn, who also serves as executive deputy editor of JAMA Network Open. But the application of AI and machine learning technologies, as described in the bill, “seems premature,” as it appears to enable the “actual prescribing of drugs, some of which are very low risk, and some of which are very high risk,” he cautioned.
Similar to autonomous vehicles being held to a higher standard than human drivers, he surmised that AI and machine learning technologies as prescribers would likely be as well. “My suspicion is that there will be a very high bar for approving these, at least initially,” he noted.
Ultimately, “I think it will come to be,” Fihn said. “And to be honest, properly developed, properly tested, properly managed, and highly regulated, it could be a good thing. But this has to be proven and shown.”
Schweikert, the bill’s sponsor, did not immediately provide additional comment to MedPage Today; however, he spoke about AI in healthcare and how it could impact government costs in an interview last year.
“Technology has to be part of the way we bend the borrowing and debt curve,” Schweikert said in the interview with Nextgov/FCWopens in a new tab or window, adding that broader adoption of AI “can make government better, faster, cheaper” in responding to Americans’ needs.
AI can improve diagnostics, Schweikert said, as well as front and back office medical operations.
“It’s here, we now just have to build the infrastructure around it,” he added, referencing a “new medical landscape.” He also noted that “technology is starting to move much faster than our regulatory rules.”
Previously, in January 2023, Schweikert introduced legislation aimed at achieving the same goal as the bill introduced earlier this month, according to Nextgov/FCW. Ultimately, though, the bill did not advance out of committee.
Meanwhile, AI in healthcare has continued to percolate as new uses emerge.
At an online panel discussion hosted by the Kaiser Family Foundation last February, health policy experts discussed the potentialopens in a new tab or window of AI to simplify prior authorization, though they added that “complete transparency” would be needed.
In many ways, AI in healthcare is still uncharted territory, and at the end of 2024 MedPage Today reported that accountability is likely to be key this year, though regulation isn’t necessarily expectedopens in a new tab or window.
In a policy position paper from the American College of Physicians last June, one of the position statements was that AI technologies should complement, not supplantopens in a new tab or window logic and decision making of physicians and other clinicians.
Regarding Schweikert’s current bill, it remains to be seen how far it may advance through the legislative process.
Five years ago, what is being proposed in the bill may have seemed like science fiction, Rodman noted. “This isn’t sci-fi,” he said. “It’s just we aren’t there yet.”