What is PD CEN ISO/TS 24971-2 ED1 - Medical devices - Guidance on the application of ISO 14971 - Part 2: Machine learning in artificial intelligence about?
PD CEN ISO/TS 24971-2 Edition 1 gives you guidance on how to apply the BS EN ISO 14971 Medical devices — Application of risk management to medical devices to machine learning-enabled medical devices.
It explains the risks that machine learning can introduce, including issues linked to training data, bias, explainability, transparency, autonomy, cybersecurity, model drift and continuous learning.
PD CEN ISO/TS 24971-2 ED1 is designed to supplement existing medical device risk management standards, so you can address machine learning-specific issues across development, testing, deployment, retraining and post-production monitoring.
The standard is intended to be used alongside BS EN ISO 14971. It also builds on the general guidance in PD CEN ISO/TR 24971 by addressing risk management challenges unique to machine learning-enabled medical devices. It provides practical guidance on topics such as bias, training and test data, explainability, transparency, model drift, continuous learning, retraining and autonomy that are not covered in the same depth by existing risk management guidance.
Who is PD CEN ISO/TS 24971-2 ED1 - Medical devices - Guidance on the application of ISO 14971 - Part 2: Machine learning in artificial intelligence for?
This publication is relevant if you develop, assess, manage, or review machine learning-enabled medical devices.
It is particularly useful for:
- Risk management professionals in medical device manufacturing.
- Regulatory affairs specialists.
- Software and AI/ML engineers.
- Clinical safety and usability experts, including human factors specialists.
- Quality assurance professionals.
- Product managers and device developers.
- Conformity assessment bodies and other stakeholders involved in regulatory review.
What does PD CEN ISO/TS 24971-2 ED1 - Medical devices - Guidance on the application of ISO 14971 - Part 2: Machine learning in artificial intelligence cover?
PD CEN ISO/TS 24971-2 ED1 covers how to adapt risk management for machine learning-enabled medical devices throughout the lifecycle.
It includes guidance on:
- machine learning-specific hazards and hazardous situations;
data quality, model design, overfitting, underfitting and non-representative training data; - transparency, explainability, usability, and human oversight;
bias, autonomy and safety characteristics that need specific attention in ML-enabled devices; - ongoing monitoring, model drift detection, retraining, updating, and rollback procedures; and
- post-production monitoring to help maintain safety and performance over time.
This publication does not apply to machine learning-enabled medical devices that use large language models or generative AI.
Why should you use PD CEN ISO/TS 24971-2 ED1 - Medical devices - Guidance on the application of ISO 14971 - Part 2: Machine learning in artificial intelligence?
If you work with machine learning-enabled medical devices, this publication helps you apply established risk management principles to technologies that behave differently from traditional devices.
Using it can help you:
- Identify and control machine learning-specific risks more clearly.
- Improve traceability and consistency in your risk management documentation.
- Align more closely with global expectations for AI and machine learning-enabled devices.
- Build safer model design, validation and data handling practices into development.
- Strengthen oversight of updates, retraining and ongoing monitoring.
- Support patient safety and device effectiveness through better lifecycle risk control.
It is among the latest international guidance documents focused specifically on applying ISO 14971 risk management principles to machine learning-enabled medical devices.