What Has Artificial Intelligence Brought to Clinical Practice? A Cross-Sectional Analysis of FDA-Authorised Devices, 2021–2026.

Abstract Background Artificial intelligence (AI) is entering defined diagnostic and decision-support tasks. The regulatory authorisation record can show the pace and distribution of this development, while selected clinical studies help explain what individual applications contribute to care. Methods We conducted a cross-sectional study of the official Food and Drug Administration (FDA) artificial intelligence-enabled device list, accessed on 6 November 2026. Unique submission identifiers were grouped by final decision year, lead panel and regulatory pathway. The primary period was 2021–2025; 2026 decisions through 29 June were reported separately. Results . The primary cohort contained 1,087 records: 129 in 2021, 162 in 2022, 226 in 2023, 235 in 2024 and 335 in 2025, a 159.7% increase. Radiology accounted for 863 records (79.4%), cardiovascular devices for 91 (8.4%) and neurology for 50 (4.6%). The 510(k)-pathway accounted for 1,055 records (97.1%), De Novo for 24 (2.2%) and premarket approval for eight (0.7%). A further 181 records had partial-year 2026 decisions. Conclusions . The FDA record documents rapid growth in listed AI-enabled authorisations, concentrated in radiology but extending across multiple specialties. Clinical contributions are task-specific and include detection and diagnostic workflow support. Authorisation establishes a regulatory basis and sustained patient benefit require separate evidence.