The agency's device center shared a discussion paper in August calling for feedback on how to ensure medical devices that use AI are safe and effective before and after they enter the market.
The challenge with generative AI is that it can provide many different outputs that may change.
This makes it more difficult to evaluate for safety and efficacy, and to ensure that a model's performance doesn't worsen over time.
The technology's tendency to produce “hallucinations,” incorrect outputs that may appear authentic, and its environmental costs are also major concerns raised by potential users and communities.
The FDA has granted breakthrough designation to a handful of medtech tools using generative AI.
The FDA has also included generative AI-based devices in a new digital health pilot, called Technology-Enabled Meaningful Patient Outcomes, or TEMPO.
【MORE】