Identifying Structural Heart Disease and Prioritizing Echocardiography Through Artificial Intelligence-Enabled Tele-ECG: An External Validation of the PROVAR+ Study.

Published in The American journal of cardiology. Abstract: Structural heart disease (SHD) is underdiagnosed where echocardiography access is limited. We evaluated the role of artificial intelligence-enabled ECG for SHD detection (AI-ECG-SHD) in cardiovascular screening. We prospectively enrolled adults undergoing public-system cardiovascular screening in two Brazilian cities. Participants underwent clinical assessment, digital 12-lead tele-ECG, and transthoracic echocardiography (TTE) interpreted by cardiologists. ECG signals were analyzed using a validated AI-ECG-SHD algorithm through an application programming interface. SHD was defined as moderate/severe ventricular dysfunction/hypertrophy, wall-motion abnormalities, moderate/severe valvular disease, congenital heart disease, or pericardial disease. AI-ECG-SHD performance was assessed at a predefined threshold and discrimination compared with a model incorporating clinical variables and major conventional ECG abnormalities. Number needed to screen (NNS) was calculated. Among 3,779 participants (median age 66 years; 56.7% women), 58.7% screened AI-ECG-SHD-positive and 24.2% had TTE-detected SHD. Major ECG abnormalities (OR=3.73), age (OR=1.22/10-years), male sex (OR=1.50), Chagas disease (OR=1.89), coronary disease (OR=3.01), and heart failure (OR=1.33) independently predicted SHD. AI-ECG-SHD probability (each 10% increase) was strongly associated with SHD (OR 1.55, 95%CI 1.51-1.61; p<0.001) and showed superior discrimination versus the conventional model (AUC 0.80, 95%CI 0.78-0.82...