Background: Artificial intelligence (AI) and machine learning are powerful computational approaches that have the capacity to automate and improve medical care delivery in orthopaedic surgery through augmentation of medical devices, from diagnostic modalities to surgical guidance. Existing research has focused on prospective device applications and ongoing clinical trials, but a comprehensive analysis on cleared devices by the FDA is lacking. Methods: A retrospective analysis was conducted for 70 FDA-cleared AI/machine learning–based medical devices (AIMDs) for orthopaedic surgery indications as of February 2025. These devices were categorized by indicated use, corresponding orthopaedic subspecialty, development history, AI architecture, and commercialization approach. For commercialization approach, active manufacturers were categorized by private or publicly traded status, acquisition history, and headquartered country. Results: Since the first orthopaedic AIMD clearance in 2017, the 3-year moving average of AIMD clearances increased from 3.0 devices/year from 2017 to 2019 to 16.6 from 2022 to 2024. Alongside this growth, deep learning emerged as the dominant AI technique, comprising 57.3% of AIMDs approved from 2022 to 2024. Spine surgery was the most common orthopaedic subspecialty for devices, representing 42.9% of devices, followed by hip and knee at 20.0%. Surgical planning predominated across orthopaedic subspecialties except in trauma, where devices focused on fracture identification and surgical guidance. 62.2% of orthopaedic AIMDs cleared from 2017 to 2019 lacked any clinical testing, but this rate declined to 19.7% from 2022 to 2024. Overall, 22.8% of orthopaedic AIMDs lacked clinical testing and 68.6% were tested with retrospective data sets. Only 8.6% were validated through a formal, prospective clinical trial. Conclusion: Although AI represents an exciting and rapidly developing area of innovation in orthopaedic surgery, improved regulatory safeguards and clinical evaluation standards are essential for the evidenced adoption and safe implementation of these promising technologies.
Lee et al. (Sun,) studied this question.