Researchers from NYU Langone Health and University of New York, both based in New York City, found discrepancies in the marketing of artificial intelligence and machine learning-enabled medical devices compared with FDA approval summaries.
The study, published July 5 in JAMA Network Open, analyzed 119 FDA 510(k) approval summaries and accompanying marketing materials for devices approved between November 2021 and March 2022. The devices were categorized as adherent, contentious and discrepant. The study found 12.61 percent of devices were discrepant, 6.72 percent were contentious and 80.67 percent were consistent between marketing and FDA clearance summaries.
Here are three other findings:
- Discrepancies between clearance documentation and marketing material was present in 20 percent of devices surveyed.
- Of radiological devices, 82.67 percent were adherent, 4 percent contentious and 13.33 percent discrepant.
- Of cardiovascular devices, 82.61 percent were adherent, 8.7 percent contentious and 8.7 percent discrepant.
Currently, there are no uniform guidelines set by the FDA to regulate artificial intelligence or machine learning-enabled medical devices and "discrepancies between FDA-approved indications for use and device marketing require articulation," the authors wrote.