The European Commission has released draft guidelines aimed at clarifying the classification of high-risk AI systems within the medtech and digital health sectors. These guidelines are particularly significant for AI technologies that utilize physiological signals, such as gait analysis or ECG readings, to assess patient health, categorizing them as high-risk under the biometric use case.
The draft guidelines, which are open for consultation until July 23, do not impose new obligations but provide a structured interpretation of existing high-risk AI classification rules. They highlight how these rules will be applied to various AI applications in medtech and pharmaceuticals. Notably, AI systems that are integral to medical devices regulated under the EU Medical Device Regulation (MDR) or In Vitro Diagnostic Regulation (IVDR) are classified as high-risk, emphasizing the importance of regulatory compliance for developers in these fields.
One key clarification in the guidelines is that the choice of conformity assessment module by manufacturers does not influence the high-risk classification. Instead, the focus is on whether the product is subject to enhanced scrutiny under relevant regulations. This distinction is particularly relevant for medtech, where most devices require third-party assessments, contrasting with other sectors where manufacturers might have more flexibility.
Additionally, the guidelines extend to AI systems used for emergency triage and patient prioritization, indicating that such tools also face high-risk classification if they operate within the medical device framework. This dual compliance requirement underscores the complexities that medtech developers must navigate as they integrate AI technologies into their products.
Overall, these draft guidelines signal a tightening of regulations around high-risk AI in healthcare, necessitating careful consideration from developers regarding compliance and the implications for patient safety and data governance.