AI medical device post-market surveillance regulations: consensus recommendations by the European Society of Radiology
Insights into Imaging, Springer Nature, 2025
Consensus recommendations from the European Society of Radiology (ESR) define how post-market surveillance (PMS) for AI-based medical devices in imaging should be organised to ensure long-term safety, effectiveness, and reliability. The guidance addresses the rapid adoption of AI in radiology and the limitations of existing PMS frameworks, which were originally designed for static medical devices and do not fully account for continuously evolving AI systems.
PMS for AI is framed as a continuous and proactive process, based on systematic collection of real-world performance data rather than passive incident reporting alone. Manufacturers are expected to predefine performance metrics, monitoring strategies, and thresholds that trigger investigation, model rollback, or retraining in response to performance drift or safety signals. Software updates, particularly for adaptive or learning systems, require strict control, documentation, and, where applicable, re-evaluation under the EU Medical Device Regulation (MDR).
Primary responsibility for PMS remains with manufacturers, including regulatory reporting of safety issues, while radiology departments contribute structured feedback and real-world outcome data. Regulators are encouraged to clarify AI-specific PMS expectations, support harmonisation across Europe, and facilitate shared registries and monitoring infrastructures.
Effective PMS depends on access to high-quality clinical data, robust GDPR-compliant data governance, and transparent documentation of AI model behaviour and changes over time. Collaboration between manufacturers, healthcare providers, regulators, and professional societies is highlighted as essential to enable safe, scalable, and trustworthy deployment of AI in radiology.