A new FDA artificial intelligence tool designed to streamline medical device reviews is faltering in its early rollout stages, raising concerns for agency staff and ethics experts about its performance, NBC News reported June 3.
Here are four notes:
- Internally known as CDRH-GPT, the AI system is still in beta testing and was developed to assist the FDA’s Center for Devices and Radiological Health in reviewing complex submissions for devices such as pacemakers and insulin pumps.
- While it has the potential to shorten lengthy review timelines, two individuals familiar with the tool told NBC News it struggles with some basic functions, including uploading documents and submitting questions.
“It still needs human supplementation,” Arthur Caplan, PhD, head of medical ethics at NYU Langone Medical Center in New York City, told the news outlet. “AI is really just not intelligent enough yet to really probe the applicant or challenge or interact.” - The tool is a part of a broader AI push within the agency led by Commissioner Marty Markary, MD, who took office April 1 and set a June 30 deadline for AI rollout across the agency. On June 2, Dr. Markary said the FDA was ahead of schedule and highlighted another tool, Elsa, that the agency has deployed to help with tasks such as summarizing adverse event data.
- Experts warned that the AI push raises questions about oversight. Richard Painter, a former government ethics lawyer who currently is a professor at Minneapolis-based University of Minnesota, said there must be clear safeguards to prevent FDA reviewers from holding financial ties to AI vendors. “A conflict of interest can greatly compromise the integrity and the reputation of a federal agency,” he said.
