The FDA announced Tuesday, April 3, that it is working on a new framework to regulate AI-based medical devices that continually learn from healthcare data.
Scott Gottlieb, MD, who announced his resignation as commissioner of the FDA last month, released a discussion paper outlining the administration’s approach to keep up with the rapid advancement of AI algorithms and ensure their safety. The agency is also asking for feedback about the paper due by June 3.
“The goal of the framework is to assure that ongoing algorithm changes follow pre-specified performance objectives and change control plans, use a validation process that ensures improvements to the performance, safety and effectiveness of the artificial intelligence software, and includes real-world monitoring of performance once the device is on the market to ensure safety and effectiveness are maintained,” Gottlieb wrote in a prepared statement from the FDA.
To date, all AI technologies cleared by the FDA have been considered “locked” algorithms, in that they don’t continually learn after each use, Gottlieb wrote in the same statement. These are modified by the manufacturer, trained on new data and the updated algorithm is then manually verified and validated.
The newly proposed framework focuses on “adaptive” algorithms that continually evolve and don’t require manual updates. These steps focus on the algorithm’s performance, manufacturer’s plan for modifications and the ability of the manufacturer to manage and control risks associated with such modifications.
Additionally, the FDA may review a software’s predetermined change control plan which would provide the agency with detailed information about the types of changes an algorithm may need based on its re-training and updated strategy and methodology used to put those changes in place.
“Artificial intelligence has helped transform industries like finance and manufacturing, and I’m confident that these technologies will have a profound and positive impact on health care,” Gottlieb added. “I can envision a world where, one day, artificial intelligence can help detect and treat challenging health problems, for example by recognizing the signs of disease well in advance of what we can do today. These tools can provide more time for intervention, identifying effective therapies and ultimately saving lives.”