The FDA has launched a pilot program using generative AI to help expedite the review of drug and device applications, part of a sweeping modernization effort by the agency.
The initiative, led by FDA Commissioner Marty Makary, MD, and Vinay Prasad, MD, director of the agency’s Center for Biologics Evaluation and Research, aims to leverage AI for a “first pass” review of application documents that often exceed 500,000 pages, according to a viewpoint article published June 10 in JAMA Network.
The FDA debuted its AI system, called Elsa, on May 8, with the large language model being intended to help prioritize inspections, summarize safety data and perform other basic review tasks, The New York Times reported June 10.
However, some agency staff have noted that the AI system has limitations, including false information being produced and character-count constraints.
In their article, Dr. Makary and Dr. Prasad highlighted the agency’s efforts to reduce reliance on animal testing and to speed up final drug approval decisions to a matter of weeks. The process of reviewing a drug can take years, according to the Times.
Dr. Makary and Dr. Prasaid also wrote that in some cases, one clinical trial could suffice instead of two.
