The two most prominent radiological societies in the U.S. are urging the federal government to proceed cautiously in its pursuit of artificial intelligence models that operate autonomously.
RSNA and the American College of Radiology spelled out their concerns in a letter sent to the Food and Drug Administration on Tuesday. In it, top officials from the two groups said they believe it’s unlikely the FDA can provide assurances of such technology’s safety in imaging care, absent further testing, surveillance and other methods of oversight.
They recommend that the FDA waits until such AI algorithms have a broader penetration into the healthcare marketplace, prior to granting its approval in the future.
“If the goal of autonomous AI is to remove the physician from the image interpretation, then the public must be assured that the algorithm will be as safe and effective as the physicians it replaces, which includes the ability to incorporate available context and identify secondary findings that would typically be identified during physician interpretation,” wrote ACR Chair Howard Fleishon, MD, and RSNA Chair Bruce Haffty, MD. “We believe this level of safety is a long way off, and while AI is poised to assist physicians in the care of their patients, autonomously functioning AI algorithms should not be implemented at this time.”
The letter comes after the agency responsible for clearing AI products held a two-day workshop in February to explore this technology’s budding role in the specialty. FDA officials, in particular, have sought to better understand how greater autonomy may impact patient safety in radiology.
The FDA reportedly plans to use feedback from the workshop to guide how it will regulate advancements in medical AI. RSNA and ACR leaders asked federal officials to focus their attention on tools that could improve care processes for both rads and patients, rather than removing the latter from the equation.
“Algorithms that assist physicians in population health management by incidentally detecting and quantifying potentially undiagnosed chronic disease would be an excellent way to begin brining autonomous AI into radiological care,” they concluded.