- FDA Commissioner Scott Gottlieb Tuesday unveiled a proposed framework to allow ongoing artificial intelligence algorithm changes based on real-world learning. Modifications to traditional software as a medical device (SaMD) that could have a significant impact on the safety or effectiveness of a device would still require a submission to FDA.
- Approved AI products to date generally have locked algorithms and do not automatically change over time as new data is collected. But Gottlieb suggests relying on periodic modifications by manufacturers may delay the promise of AI to actively learn and potentially improve intervention timeliness and outcomes.
- The idea, laid out in a discussion paper, is to determine what type of AI/machine learning-based SaMD modifications, if any, could potentially be exempted from premarket submission requirements. FDA is formally asking for feedback in a request for information on the discussion paper by June 3.
The paper is part of the agency's evolution to adapt current rules to the emerging technologies.
In the framework, FDA argues a total product lifecycle approach, including performance monitoring, is needed to regulate AI/ML SaMD with reasonable assurance of safety and effectiveness of a product.
"This first step in developing our approach outlines information specific to devices that include artificial intelligence algorithms that make real-world modifications that the agency might require for premarket review. They include the algorithm's performance, the manufacturer's plan for modifications and the ability of the manufacturer to manage and control risks of the modifications," Gottlieb said in a statement.
The agency is proposing a "predetermined change control plan" may be needed to provide information to FDA about what anticipated changes an algorithm may undergo, along with an explanation of the method used to implement those changes.
"We also anticipate that in certain cases, the SaMD's risk or the intended use may significantly change after learning," FDA's white paper states. The agency warns that such a change would trigger a need for a new premarket submission.
Epstein Becker Green device attorney Brad Thompson points out FDA admits in the discussion paper that additional statutory authority may be needed to fully implement such an idea.
"I am worried a bit that FDA is becoming really good at coming up with ideas and not so good about carrying through with them," Thompson told MedTech Dive in an email. "It's unclear whether they are saying that certain portions of this idea can be implemented right away through guidance, while others may have to await legislation."
FDA plans to issue draft guidance based on the input it receives on the new discussion paper using its "current authorities in new ways to keep up with the rapid pace of innovation and ensure the safety of these devices," according to Gottlieb.
While Thompson remains concerned how both the new AI framework and FDA's Software Precertification Program will be implemented, he praised FDA for its pursuit of innovative regulatory approaches for new technologies like AI.
"I just hope that the agency can carry through with some of these new initiatives all the way to completion," Thompson said. "Between the two programs – precertification and this new AI initiative – speaking personally I think I'm more excited about the new AI initiative."
In December 2017, FDA released a slate of draft and final guidance outlining how it planned to approach regulating SaMD and certain types of clinical decision support software.
The idea for an AI framework may have originated within industry. In response to slate of regulatory actions, AdvaMed called for FDA to clarify when certain clinical decision software should not be regulated in a written comment.
The medical device lobby said it is pleased with what may be Gottlieb's final SaMD action before leaving the agency, expected to happen Friday.
"We're still reviewing the discussion paper, but overall we're pleased to see FDA begin the discussion about how it will regulate artificial intelligence and we look forward to working with them," AdvaMed spokesperson Mark Brager told MedTech Dive.
One question that remains is if FDA has the resources to apply a total product lifecycle approach to regulation. During the last round of medical device user fee negotiations, it was a sticking point between the agency and industry. When asked if FDA is now well-resourced enough to realize this vision for AI/ML SaMD, the agency declined to comment.
"At this time, the discussion paper is meant to solicit feedback on our proposed approach. It's a first, foundational step," the spokesperson told MedTech Dive.