Few FDA Approved AI/ML Orthopaedic Devices Have EU MDR Equivalents or Peer-Reviewed Validation

Area of Science:

• Orthopaedic medical devices
• Artificial intelligence (AI) and machine learning (ML)
• Regulatory science

Background:

• AI/ML integration in orthopaedics offers enhanced surgical planning and personalized care.
• Existing US and EU regulatory pathways face challenges with AI/ML medical devices.
• The EU Artificial Intelligence Act provides a framework, but the US lacks a specific directive.

Purpose of the Study:

• Assess FDA-approved AI/ML orthopaedic devices for EU approval (MDR).
• Determine the frequency of FDA 510(k) approvals using non-AI predicate devices.
• Evaluate the availability of independent, peer-reviewed evidence for AI/ML orthopaedic devices.
• Examine the disclosure of training and validation dataset details in regulatory documents.

Main Methods:

• Analyzed 37 AI/ML-enabled orthopaedic devices from FDA data (May 2025).
• Assessed EU availability via EUDAMED and manufacturer websites.
• Reviewed PubMed and manufacturer sites for peer-reviewed evidence.
• Examined regulatory documents for predicate devices and dataset disclosures.

Main Results:

• 38% of devices were approved in the EU; most were available in the US first.
• 62% of FDA-approved devices used non-AI predicate devices.
• Peer-reviewed evidence was found for 30% of devices, often with manufacturer involvement.
• Dataset details were disclosed in 57% of 510(k) summaries.

Conclusions:

• AI/ML orthopaedic devices are entering markets under current regulations, but at different paces in the US and EU.
• Reliance on non-AI comparators and limited postmarket evaluation pose concerns for long-term clinical reliability.
• Future research should compare recall rates across jurisdictions to evaluate AI-specific regulatory frameworks' impact on safety and reliability.