Clinical AI Certification Center Launched to Accelerate Medical Device Approval

Details

The Taiwan MOHW established the 'Clinical AI Certification Center' in response to the rapid adoption of AI technology in healthcare. This center is pioneering a cross-hospital external validation system designed to test whether AI medical devices function effectively in real clinical settings. The core problem addressed was that AI models often suffered performance degradation and bias because their training data samples were limited to single institutions or specific demographics, lacking population representativeness. To solve this, the MOHW Information Department partnered with the TFDA and multiple academic hospitals (e.g., Taichung Veterans General Hospital) to build a validation system. The center utilizes two key approaches: 'Centralized Data Validation' and 'Federated Learning Platform Validation.' This structure allows developers to test model performance across multiple hospitals while maintaining patient privacy. The center issues standardized reports based on TFDA regulations. Crucially, the four centers have also established FHIR (Fast Healthcare Interoperability Resources) servers, creating Taiwan's first cross-hospital electronic medical record database based on this standard. This significantly enhances data interoperability and standardization, boosting the practical usability and scalability of AI models in clinical settings.