China NMPA released the "Clinical Evaluation Guideline on AI-assisted Detection Medical Devices (Software)" on November 7, 2023. Based on FDA guidance on "Computer-Assisted Detection Devices Applied to Radiology Images and Radiology Device Data" and "Clinical Performance Assessment: Considerations for Computer-Assisted Detection Devices Applied to Radiology Images and Radiology Device Data", the scope of the guideline includes Class III AI-assisted detection devices utilizing AI algorithms for pattern recognition in diagnosing conditions like lung nodules, breast nodules, fractures, vascular narrowing, and colon polyps. These devices may function as Software as a Medical Device (SaMD) or embedded software.
Clinical trials for these devices prioritize enhancing disease detection accuracy, involving controlled designs such as randomized parallel, crossover, or multiple reader multiple case (MRMC) designs. Radiologists are often included as study subjects to address reader variability, and real-world data from clinical practice informs imaging sample collection.
Primary endpoints in trials include sensitivity, specificity, and ROC measures aligned with product design characteristics. The guideline underscores transparent sample size calculations, statistical analysis, and proper reader training to minimize bias. Quality control measures for image review, including blind review and diverse reader representation, are recommended.
Non-diagnostic functions, such as structured report generation and workflow optimization, can be assessed for safety and effectiveness. The guideline emphasizes proper evaluation to ensure reliability and performance. Instructions for Use (IFU) recommendations include a clinical trial summary and clear product scope definition with warnings for safe use and awareness of guideline updates.
An annex provides examples of AI-assisted detection products for colon polyps and pulmonary nodules, concluding with warnings about lesion boundary segmentation accuracy and potential risks tied to guideline changes over time.
I am interested in your feedback on your experience in using the FDA equivalent guidelines mentioned above.
------------------------------
Grace Fu
China Med Device, LLC
MA, U.S.
gpalma@ChinaMedDevice.com978-390-4453
www.ChinaMedDevice.com------------------------------