FDA-Cleared AI Medical Devices — Less Than 1% Tested for Clinical Benefit (2026 Evidence Review)
OVERVIEWA growing body of reporting confirms a significant evidentiary gap in the FDA's AI medical device clearance landscape: of 1,357 FDA-cleared AI medical devices, fewer than 1% have been tested for clinical benefits to patients, and only three have been tested on patient outcomes specifically. These figures, reported across multiple outlets including Inside Precision Medicine, TechTarget, News-Medical, and AuntMinnie, draw on the same underlying research but represent consistent findings across independent coverage. The data raises fundamental questions about the FDA's clearance process for AI-enabled devices. Clearance through the 510(k) pathway — which requires demonstrating substantial equivalence to a predicate device rather than clinical superiority — does not mandate patient outcome evidence. AI devices cleared via this route may demonstrate technical performance on benchmark datasets without ever being tested in real clinical environments with actual patient populations. The implications extend beyond regulatory process: health systems deploying these devices, clinicians relying on their outputs, and patients subject to AI-assisted decisions are operating with limited evidence that the tools produce better health outcomes than existing alternatives. This evidentiary gap is particularly acute in radiology, where the majority of FDA-cleared AI devices are concentrated. Ongoing policy discussions at the FDA, including the TEMPO pilot program for generative AI devices and…
Subscribe to read full wiki analysis
Full wiki content — including source provenance, contradiction tracking, related pages, and analyst notes — is available to Subscriber members.
View Plans & SubscribeNew users get a 7-day trial — no credit card required.