Standards Development

Scope

This document extends the framework of post-market surveillance for machine learning-enabled medical devices (MLMD).

This document describes special requirement, metrics, methods and procedure to implement post -market surveillance for MLMD.

Purpose

The purpose of this standard is to clarify specific requirement and procedure of post-market surveillance for MLMD, facilitating post-market activities in regulation and quality management.

The MLMD industry is developing rapidly all over the world. Thousands of products have been approved and applied in healthcare. Innovative and expanding use of machine learning techniques is met with rising public concern on its impact on product quality. For example, is the MLMD application as accurate and reliable in the real world as claimed by the manufacturer? Is there underlying algorithm discrimination for different groups of patients?

To protect the well-being of stakeholders and promote post-market regulation, it is important to understand how MLMD behaves in the post-market phase and whether it continuously ensures safety and effectiveness. A framework to guide post-market surveillance of MLMD is thus needed. Technical considerations include but are not limited to documentation, performance metrics, methodology, human -machine interaction, feedback collection and handling. Ethical consideration, such as algorithm fairness and transparen cy, also needs to be covered. This project will promote approximate collection and analysis of post -market information by stakeholders connected to the lifecycle of MLMD, such as manufacturers, regulators and distributors.