The FDA has outlined a proposed framework for regulating AI-based software used as a medical device that can continue learning and adapting after reaching the market.
The framework centers on what the agency calls a “predetermined change control plan,” which would allow manufacturers to submit an upfront plan describing how an algorithm is expected to change over time instead of filing a new premarket submission for every update, according to an FDA discussion paper.
The plan includes two components: “SaMD pre-specifications,” which describe the types of modifications a manufacturer anticipates, and an “algorithm change protocol,” which details how those modifications would be made safely. Modifications that fall within an approved plan could be documented and implemented without additional FDA review. Changes outside the plan, such as a shift in intended use, would still require a new premarket submission.
The discussion paper builds on a series of steps the FDA has taken toward regulating AI-enabled devices. A 2023 draft guidance on machine learning-enabled device modifications recommended that manufacturers account for race, ethnicity, disease severity, gender, age and geography in the data used to train and validate their algorithms. A broader draft guidance followed in January 2025, covering the full device lifecycle from design and development through postmarket monitoring. It was the agency’s first guidance to address the entire lifecycle of AI-enabled devices and came as the number of such devices authorized by the FDA surpassed 1,000.
At the Becker's 11th Annual IT + Revenue Cycle Conference: The Future of AI & Digital Health, taking place September 14–17 in Chicago, healthcare executives and digital leaders from across the country will come together to explore how AI, interoperability, cybersecurity, and revenue cycle innovation are transforming care delivery, strengthening financial performance, and driving the next era of digital health. Apply for complimentary registration now.
