Software failures in radiology devices: twenty years of U.S. FDA medical device recalls, 2005-2025

Abstract

Software is now part of most medical devices, and its errors can harm patients. This study uses the public openFDA database, which lists 20,576 U.S. medical device recall events initiated in 2005-2025, to ask how often recalls are caused by software in radiology compared with other medical specialties, and whether this has changed as artificial intelligence has entered radiology. Among recall events with a determined root cause, software was the cause of 38.1% of radiology recalls in 2005-2011, 47.9% in 2012-2018 and 40.0% in 2019-2025, against 6.0%, 9.4% and 11.4% for other devices. Radiology accounted for 16.9% of these recalls but 49.6% of software recalls. The results support the author's expectation that software fails more often in radiology, but not her expectation that the spread of artificial intelligence would increase software recalls: the FDA authorized 1,013 AI-enabled radiology devices in 2019-2025, against 82 before, while the software share of radiology recalls did not rise (28.9-44.6% a year in 2019-2025) and the gap with other specialties narrowed.

Keywords: medical devices, software, recalls, radiology, FDA, artificial intelligence

Әдебиеттер тізімі

1. Simone L.K. Software-related recalls: an analysis of records // Biomedical Instrumentation & Technology. 2013. Vol. 47, No. 6. P. 514-522. https://doi.org/10.2345/0899-8205-47.6.514
2. U.S. Food and Drug Administration. openFDA: device recall data (bulk download, export of 27 October 2026). https://open.fda.gov/apis/device/recall/ (accessed 29.10.2026).
3. U.S. Food and Drug Administration. Artificial Intelligence-Enabled Medical Devices (list, CSV). https://www.fda.gov/medical-devices/artificial-intelligence-enabled-medical-devices/list-artificial-intelligence-enabled-medical-devices (accessed 30.10.2026).

Дәйексөз келтіру үлгісі

Zhumabek Sayana Software failures in radiology devices: twenty years of U.S. FDA medical device recalls, 2005-2025 // Central Asian Scientific Journal. – 2026. – No. 4(32), Vol. 2. – Pp. 236–239.