I am a pediatric cardiologist and have cared for children with heart disease for the past three decades. In addition, I have an educational background in business and finance as well as healthcare administration and global health – I gained a Masters Degree in Public Health from UCLA and taught Global Health there after I completed the program.
“By law and institutional culture, FDA seeks to apply a single tool to its approach to regulation- the requirement for pre-market approval. The agency is accustomed to compelling innovators to submit evidence to FDA and seek permission for marketing.”
Scott Gottlieb, former FDA commissioner
Many AI/ML algorithms have been published but very few have reached the gates of FDA and received approval. These authors from Europe reviewed the currently available AI/ML-based medical devices and algorithms that have been approved by the U.S. Food and Drug Administration (FDA).
In Europe, the European Medicine Agency (EMA) is the regulatory body for these devices. This 2020 article is still useful to have as a reference for a snapshot of the regulatory landscape of the FDA for AI/ML devices. The authors claimed to have the first comprehensive and open access database of FDA approved AI/ML devices since 2012.
There are in total 64 AI/ML-based and FDA approved medical devices and algorithms. Only 29 (45%) of these approved devices, however, mentioned any AI/ML elements in the official FDA announcement. The 510(c) clearance mechanism was the approval mechanism for a majority of 55 (85.9%) of these devices. This clearance is based on whether the algorithm is shown to be at least as safe and effective as another legally marketed algorithm. An additional 8 (12.5%) were approved via the de novo pathway clearance route while only 1 (1.6%) received premarket approval (PMA) clearance.
The de novo pathway is used to classify those novel medical devices for which there are no legally marketed counterparts but have adequate safety and effectiveness. Lastly, the aforementioned PMA clearance is given to Class III medical devices as these have a larger impact on human health and thus require a more thorough scientific and regulatory process to determine their safety and effectiveness.
The authors also commented that the subspecialty areas of radiology, cardiology, and internal medicine/general practice had 30 (46.9%), 16 (25%), and 10 (15.6%) of these devices. The regulatory process at the FDA deserves credit for being more flexible than the past decade, but is still lacking in efficiency in some regards when it comes to AI/ML devices.
This is the conundrum: how to advance AI/ML tools with appropriate expediency while retaining stringent regulatory scrutiny. The pandemic has demonstrated that there are times that expediency will (and should) outweigh a rigorous regulatory process as lives need to be saved. The basic issue at hand is that the technology is moving at an exponential trajectory while the regulatory process is much slower at a more linear trajectory.
The full article can be read here