Healthcare AI devices lacking publicly available clinical study information may be more likely to be recalled, according to a June 11 study published in JAMA Network Open. Researchers said the findings emphasize the importance of clinical validation and postmarket monitoring for AI-enabled medical devices.
Here are four notes from the study:
- Researchers examined 903 FDA-authorized AI-enabled medical devices approved between 1995 and 2024. Of those, 43 devices, or 4.8%, were recalled after a median of 458 days.
- Devices with missing information on supporting clinical studies were associated with a higher hazard of recall than devices backed by published clinical studies.
- Devices flagged in the CORE-MD postmarket surveillance tool and those identified in multiple surveillance databases showed higher recall risk than devices without reported safety signals.
- Incorrect use of an AI-enabled device was associated with a higher recall hazard. Researchers also found software-related problems and concerns involving device outputs were among the most frequently reported issues.
