TrialSite News recently attended and reported on the Utah Life Sciences Summit held earlier this November at the Salt Lake Marriot. During which, three tracks were provided for registrants to learn more on a range of subjects relating to the life sciences industry. In the first track, TrialSite News attended the Life Sciences session, titled “Trending Topics at FDA: De Novo Classification Proposed Rule & AI/ML Software Regulation.”
The presenter of this session was Seth Olson, an associate attorney in Washington D.C. at Hogan Lovells Medical Device and Technology Group, which has offices across the globe. In this role, Olson provides general counsel to companies regulated by the FDA. He previously served as regulatory counsel at the FDA’s Center for Devices and Radiological Health, where he advised on agency enforcement actions, provided regulatory counsel regarding FDA premarket reviews, responded to 513(g) requests for information, participated in policymaking working groups, and aided in the development of FDA guidance and administrative orders.
De Novo Classification & Proposed Rule
Olson covered the FDA’s De Novo Premarket Classification, its history, and its comparison to a PMA or a 510(K). He informed the audience of the FDA’s proposed rule to establish regulations for the De Novo classification process, published in December of 2018, as well as the components of a De Novo Classification Request. De Novo classifications are up and coming—even companies like Bose® and 23andMe have De Novo classifications.
He also described a De Novo classification as a way to reclassify Class 3 devices to a Class 1 or Class 2, and conveyed how it is intended for devices where no predict device exists. During a De Novo classification, the FDA will identify the potential risks, provide general controls, and come up with special controls (such as labeling or post-market surveillance studies) for a medium-to-low-risk device differentiated from any other product currently on market.
Fees over 100K were added in 2017 to all De Novo applications, hence the spike in De Novo applications in 2018 when the fee was waived. In addition, Olson informs that many De Novo classification regulations are not yet found in the CFR, but can be found online here.
AI/ML Software Regulation
Audience members learn from Olson that the FDA interestingly uses John McCarthy’s definition of AI: “the science and engineering of making intelligent machines, especially intelligent computer programs.”
In the 21st Century Cures Act, signed into law in 2016, new authority is provided to the FDA, allowing them to recruit and retain scientific, technical, and professional experts. It also establishes new expedited product development programs.
The term Software as a Medical Device(SaMD) used by the FDA is defined by the International Medical Device Regulators Forum (IMDRF) as “software intended to be used for one or more medical purposes that perform these purposes without being part of a hardware medical device.”
Under the current regulatory framework, there are two types of AI/ML SaMD as determined by the FDA:
- AI algorithms that are locked prior to marketing
- AI algorithms that can adapt over time based on real-world evidence
Olson explains that the FDA’s current device regulations are not designed for devices that are continually adapting, meaning every time your device makes a change, you’re required to inform and receive approval from the FDA. This makes it extremely burdensome for companies and the FDA, who does not have the capacity to review that many submissions
For those interested in learning more about the FDA’s regulations of Artificial Intelligence/Machine Learning (AI/ML), please see here.
Call to Action: Keep your eye out for future articles covering more from the Utah Life Sciences Summit, or receive them directly by signing up for TrialSite News’ Daily Digest.