Summary - F.781.2 (06/2024) - Quality assessment requirements for artificial intelligence/machine learning-based software as a medical device

With the advent of artificial intelligence/machine learning (AI/ML) and its strength in faster and more accurate disease detection and diagnosis, more timely and widespread adoption of decision-making assistant (DMA) software as a medical device (SaMD) could improve health for human beings. However, this does not mean the AI/ML-based DMA-SaMD for decision-making is ready for the clinic; AI/ML technology can only be used with complete confidence if it has been quality controlled through a rigorous evaluation in a standardized way. Performance and usability shall be assessed under a reliable and rigorous evaluation with a robust method to substantiate AI/ML-based DMA-SaMD quality.
Recommendation ITU-T F.781.2 provides a requirement framework for the quality assessment with a perspective of lifecycle management for AI/ML-based DMA-SaMD. It describes the quality assessment principles and process in the life cycle of AI/ML-based DMA-SaMD, including requirement analysis, data collection, algorithm design, verification and validation, change control and other stages when using AI/ML technology to assist medical staff in making clinical decisions by providing suggestions on diagnostic and treatment activities.