FDA published a final order on September 11 creating 21 CFR 870.2380, a Class II regulation for cardiovascular machine learning-based notification software. The order took effect the same day. The classification applies back to August 3, 2023, the day FDA granted Viz.ai's De Novo request for Viz HCM.

The identification language is broad. FDA defines the type as software that employs machine learning techniques to suggest the likelihood of a cardiovascular disease or condition for further referral or diagnostic follow-up. Viz.ai filed the De Novo request on January 10, 2023, so the class took just over three years to reach the CFR.

Four special controls come with the class: clinical performance testing, software verification and validation with hazard analysis, a human factors assessment, and labeling. The clinical control is the one with teeth. It calls for independent test datasets drawn from geographically diverse sites. The labeling control requires a warning against over-reliance on the output.

The practical effect is a predicate. A company whose software flags a cardiovascular condition from non-invasive physiological inputs can now file a 510(k) against this regulation instead of assembling a De Novo request. Read the special controls as the test plan, because that is what a reviewer will hold the submission against.