FDA classifies cardiovascular machine-learning notification software as Class II device

# What the FDA decided

On September 11, 2026, the Food and Drug Administration issued a final order classifying "cardiovascular machine learning-based notification software" as a Class II medical device subject to special controls. The underlying classification has been applied since August 3, 2023, but the final order formalizes requirements and the regulatory pathway for this category of software.

# What this software does and does not do

The FDA defines this software as systems that use machine-learning techniques to suggest the likelihood that a patient has a cardiovascular disease or specific condition. These tools analyze one or more noninvasive physiological inputs collected during routine care and identify a single condition that may warrant further diagnostic follow-up or referral.

# Why the FDA placed it in Class II

The special controls address risks such as false positives and false negatives that could contribute to incorrect diagnoses or treatment choices. Required safeguards include clinical performance testing, nonclinical performance testing, and appropriate labeling that explains intended use and limitations.

# The Viz HCM case that triggered the classification

The regulatory action began with Viz.ai Inc.'s De Novo request for Viz HCM, submitted January 10, 2023. Viz HCM is designed to flag electrocardiogram (ECG) findings that could suggest hypertrophic cardiomyopathy (HCM), a condition in which the heart muscle thickens.

# What this means for patients and manufacturers

For patients: an AI alert may provide clinicians with another reason to investigate a possible heart condition, but it does not replace clinician diagnosis or diagnostic testing. The software's purpose is to flag possible conditions so clinicians can decide on further testing.

For manufacturers: by creating a defined Class II category via De Novo, the FDA also created a potential predicate device. Future developers of substantially equivalent cardiovascular machine-learning notification software may be able to pursue the 510(k) premarket notification pathway instead of repeating the De Novo process or going through premarket approval. That can reduce regulatory hurdles for bringing similar products to market.

# Broader regulatory context

The decision fits within a larger push to define regulatory approaches for machine-learning and AI-driven medical tools while balancing innovation and patient safety. The special controls required by the FDA target predictable technical and clinical risks for notification software and specify testing and labeling expectations manufacturers must meet.