The U.S. Food and Drug Administration's Regulatory Framework for AI-Enabled Medical Devices

The U.S. Food and Drug Administration (FDA) implements a risk-focused oversight mechanism for the classification, validation, and market authorization of medical devices utilizing artificial intelligence and machine learning (AI/ML) technologies.

The Center for Devices and Radiological Health (CDRH) and the Digital Health Center of Excellence (DHCoE) within the FDA manage the evaluation of AI-enabled software as a medical device (SaMD) or software in a medical device (SiMD). The agency distinguishes between medical devices and non-device clinical decision support systems based on four legal criteria.

Market entry processes are divided into three main pathways based on risk level: the 510(k) pathway for moderate-risk devices (more than 95% of all listed devices), the De Novo pathway for novel and low-to-moderate risk devices (3-4%), and the PMA pathway for high-risk Class III devices (less than 1%).

In October 2021, the 10 Good Machine Learning Practice (GMLP) principles published by the FDA, Health Canada, and the UK's MHRA established quality standards throughout the product lifecycle. According to the FDA's public database, more than 75% of approved AI-enabled devices are concentrated in the field of radiology.