FDA’s List of AI and Machine Learning Devices is an Invaluable Gift (Thursday, October 6, 2022)
For developers of medical devices using artificial intelligence (AI) and/or machine learning (ML) tools, precedent information is critical. Being a new field, there are few public resources to plan and strategize development steps for FDA approval. FDA’s list of AI/ML-enabled medical devices is hence a priceless gift to the developers. FDA has been maintaining this list for a few years, but it is sporadically updated, and its presence is not widely publicized. This week FDA added 178 new devices to this list. The list can be downloaded as an Excel worksheet, but the best experience of this list is on FDA’s website because that’s where all the links to the individual applications reside. Once can search the list for relevant predicate devices and research them for useful information. This list settles one question for all; FDA prefers the 510k pathway for AI/ML-enabled medical devices. Of the 521 devices included on the list, only three were approved via the PMA pathway, all the rest were cleared as 510K devices. Only 18 devices, or about 3% of the devices on the list, required the denovo 510k pathway, of which 16 were in the last 5 years. Interestingly, the list includes a device cleared in 1995 as a PMA device, much before the terms AI and ML were probably used for medical devices, indicating that FDA has a broader definition of AI/ML-enabled devices than the industry. About 450 of the 521 devices (more than 85%) on the list were cleared/approved in the last 4 years. The most useful aspect of this list is that it includes links to the 510k summaries for each device on the list which in turn include a summary of what tests, studies and other information were generated to support those applications. While it is possible to search for devises cleared by FDA directly, there is always a risk that one would miss a device cleared by FDA or not know the specific classification of the device by FDA. This list addresses those issues. This an invaluable gift to AI/ML-enabled medical device developers, indeed.
Dr. Mukesh Kumar
Founder & CEO, FDAMap
Linkedin: Mukesh Kumar, PhD, RAC