External validation of artificial intelligence-guided electrocardiogram (AI-ECG) in detecting asymptomatic left...

Asymptomatic left ventricular dysfunction (LVD) represents a key and frequently under-recognized stage in the heart failure (HF) continuum. Early identification enables targeted prevention and timely initiation of evidence-based therapies. Artificial intelligence-enabled electrocardiography (AI-ECG) may provide a scalable alternative to imaging for LVD detection. This systematic review synthesized evidence from exclusively externally validated AI-ECG studies evaluating the detection of asymptomatic LVD, with symptomatic HF cohorts included to contextualize the evidence base across the HF continuum.

A systematic literature search was conducted on 8th September 2025. The main output was the pooled area under the receiver operating characteristic curve (AUROC).

From 3,587 records, 12 asymptomatic and 30 symptomatic studies were identified (538,735 participants). In asymptomatic cohorts evaluating left ventricular systolic dysfunction (LVSD), AI-ECG demonstrated promising but heterogeneous discriminatory performance (I²>85%) with pooled AUROCs of 0.87 (95% CI 0.75-0.99) for LVEF ≤35%, 0.93 (95% CI 0.88-0.97) for LVEF ≤40%, and 0.89 (95% CI 0.85-0.97) for LVEF ≤50%. Symptomatic cohorts showed AUROCs of 0.89 (95% CI 0.86-0.92), 0.90 (95% CI 0.88-0.92), and 0.85 (95% CI 0.83-0.86) respectively. Asymptomatic left ventricular diastolic dysfunction (LVDD) demonstrated moderate performance (AUROC 0.76, 95% CI 0.66-0.86) with no externally validated symptomatic LVDD studies identified. Small-study effects were detected exclusively in asymptomatic cohorts.

Discriminatory performance of AI-ECG in asymptomatic LVSD is broadly comparable to symptomatic populations. High risk of bias driven by prolonged ECG-to-echocardiogram intervals reflecting real-world logistical constraints necessitate cautious interpretation of results and further prospective validation.