FDA Cleared AI Devices Lack Outcome Testing

A study published in PLOS on August 19, 2026, analyzed all 1,357 AI-based medical devices authorized by the U.S. United States — Food and Drug Administration (FDA) as of December 5, 2025. The researchers, led by Rawan Abulibdeh of the University of Toronto, found that only 34 devices had been included in registered clinical trials, with results posted for 12 and peer-reviewed manuscripts published for 12. Only three devices had been tested on patient-centered outcomes such as death rates, strokes, hospitalizations, and quality of life. Most studies were conducted in highly resourced healthcare systems and excluded key patient subgroups, including pregnant women, adults over 75, and non-English speakers. The authors highlight that the FDA's 510(k) pathway allows devices to be cleared by demonstrating 'substantial equivalence' to existing devices, without requiring evidence of improved patient outcomes. They propose a three-phase framework for stronger clinical validation, including prospective studies in diverse settings and post-clearance multicenter trials. The study also warns that under-validated AI tools could be adopted in low- and middle-income countries, potentially amplifying healthcare disparities.