Regulatory Focus™ > News Articles > 2022 > 6 > FDA should step in to help spur AI/ML standards: Expert

 

FDA should step in to help spur AI/ML standards: Expert

Bradley Thompson/ Epstein, Becker and Green

Regulatory NewsRegulatory News
Posted 13 June 2022 | By Ferdous Al-Faruque 

FDA should step in to help spur AI/ML standards: Expert

2863 The US Food and Drug Administration (FDA) should support the burgeoning integration of artificial intelligence and machine learning (AI/ML) in medical devices by encouraging the development of new standards and setting deadlines for their implementation.
 
That’s according to Brad Thompson, a regulatory attorney at Epstein, Becker and Green. Thompson spoke with Regulatory Focus on the state of AI/ML and what can be done to help move the regulatory needle forward. While he supports organizations coming up with standards on their own, he says regulators should speed things up.
 
In the absence of clear expectation from the FDA, organizations such as Xavier University and private industry have stepped in to  fill the gap with standards that address issues such as data requirements  and integrity, Thompson said. On the government level, the National Institute of Standards and Technology is trying to do the same, he added.
 
To date, the FDA has relied on its 20 years of experience with imaging data to direct sponsors to guidances to understand data requirement needs – but that’s not enough, he said.
 
“I think FDA should encourage... standards organizations and industry bodies to really try and develop best practices that FDA can then incorporate by reference into their guidance documents, into their regulatory oversight,” said Thompson. “My concern with [the current] approach... is standards groups are notorious for how long they take to develop standards. They can easily spend years debating things before they get to a consensus decision.
 
“So maybe FDA has to put deadlines around the process to say, 'Look, we're going to decide what to do on this topic after this date and if you're a standards group and you want to have input, then get going and have something for us by that date, and then we'll decide,' " he added.
 
While FDA in recent years has been stretched thin in managing the COVID-19 pandemic, Thompson said the agency does have policy staff with sophisticated understanding of AI/ML algorithms to issue good policy. The lack of bandwidth is really in the review areas, he said.
 
In recent years the FDA has argued in front of Congress for policy changes that would give it explicit authority to allow predetermined change protocols and the creation of a Pre-Certification (Pre-Cert) program that would let new digital health products on the market based on the trust regulators have built with specific manufacturers.
 
 
FDA already has the authority it needs to allow AI/ML products on the market through the de novo pathway, where sponsors allow their algorithm to change within certain pre-specified parameters, according to Thompson.
 
The FDA has also argued it needs the authority to allow additional data collection and monitoring in the postmarket setting, but Thompson said that he believes that would be overly intrusive.
 
“I don't think the FDA can handle the volume of data and I don't think, from a judgment standpoint, we can count on it not to overreact to some of that data. I think it just becomes a huge bottleneck,” said Thompson. “I think the [current] system that requiring manufacturers to alert the FDA of adverse events continues to be a reasonable approach.”
 
“It sounds very Big Brotherish to say that the FDA wants authority to watch all medical devices directly as they perform and whatever signals they send off,” he added. “
 
In terms of the FDA’s planned Pre-Cert Program, Thompson has long made it clear he’s not a fan. He says the agency is right that it needs new authority to implement company-based oversight instead of the traditional product-based oversight. But he said it’s a sure recipe to create an ecosystem where established companies will have an unfair advantage over newcomers.
 
“It basically entrenches industry so that if you invest heavily in having a good relationship with the FDA such that you get this privileged position, newcomers are at a very great disadvantage,” said Thompson. “And the medical device industry, in contrast to the drug industry, depends mightily on new companies entering all the time... They would be at a huge disadvantage if we lived in a world where Pre-Cert existed. I just think there are all sorts of things that they can do within the existing statute, and... I can't see any reason to go to Congress right now for new legislative authority.”

 

© 2022 Regulatory Affairs Professionals Society.

Discover more of what matters to you

9;11;18;20;31;