AI-Enabled Smartwatch Detection of Structural Heart Disease
11/04/2025
In a preliminary study being presented at the 2025 American Heart Association Scientific Sessions, an AI algorithm tested on smartwatch ECGs accurately identified structural heart disease, detecting reduced ventricular function, valvular lesions, and hypertrophy.
Echocardiography remains the diagnostic standard but is constrained by equipment, personnel, and episodic availability. Consumer wearables, by contrast, can flag abnormalities continuously and at scale, shifting case finding earlier and altering referral patterns and resource needs.
The developers trained the model on 266,054 ECGs and tested it prospectively in about 600 adults. In that real-world cohort, the model showed ~88% accuracy, 86% sensitivity, and a 99% negative predictive value.
Next steps include further assessing the tool and its potential for integration in community-based preventive screening services.