Medical Devices – Latest Updates

While “manufacturer” used in the MDR/ IVDR is synonymous with “provider” in the AIA, “deployer” in the AIA does not correspond to “user” in the MDR/ IVDR. The MDR and IVDR address risks associated with medical device software, but do not cover specific risks that could arise from artificial intelligence (AI). I these cases, the AIA, which addresses risks and challenges in the use of AI, is complementary.

To avoid redundant documentation while ensuring compliance with all requirements, MDAI manufacturers have the option to integrate testing and reporting procedures as well as AI-related information and documentation into existing MDR/IVDR procedures in accordance with Article 8(2) of the AIA.

According to the AIA, an MDAI is considered a high-risk AI system if it is a safety component or a medical device, or if a notified body is involved in the conformity assessment. However, the high-risk classification according to the AIA does not affect the risk classification under the MDR/IVDR.

The document also contains questions and answers about the requirements for management systems, data governance, technical documentation, transparency and human oversight. It provides explanations on the accuracy, robustness and cybersecurity that MDAI manufacturers must meet. Clinical evaluation/performance evaluation and testing must be conducted in accordance with the MDR/IVDR and the AIA, and the document explains which conformity assessment procedures apply to AI systems. It also covers the procedures for significant changes to the AI systems and post-market monitoring. The key difference in the latter case compared to the previous procedure is the need to monitor interactions with other AI systems.

All in all, the document contains the first round of the most frequently asked questions and will continuously be revised and updated with new questions and answers to meet the ongoing need for clarity in the field of medical device AI.