AI Enabled Single Lead Wearable Electrocardiogram (ECG) Screening for Early Detection of Left Ventricular Systolic Dysfunction (LVSD)

Abstract

Left ventricular systolic dysfunction (LVSD) is an important precursor to heart failure and may remain clinically silent until ventricular impairment has progressed. Although echocardiography remains the standard confirmatory method for assessing left ventricular ejection fraction, its use as a broad screening tool is limited by cost, infrastructure requirements, specialist availability, and delayed referral pathways. This paper examines the potential of artificial intelligence-enabled single-lead wearable electrocardiogram screening for early LVSD risk detection. The study proposes a diagnostic screening framework in which wearable ECG signals are acquired, preprocessed, segmented, and analyzed using an AI-based classification model to generate an LVSD probability score. In the simulated evaluation, LVSD was defined as LVEF ≤ 40%, and the proposed model achieved an accuracy of 89.0%, sensitivity of 87.5%, specificity of 89.3%, F1-score of 71.8%, negative predictive value of 97.4%, and AUROC of 0.91. These findings suggest that AI-enabled single-lead wearable ECG may be clinically useful for identifying high-risk individuals and ruling out low-risk cases before referral for confirmatory imaging. However, the system should not be interpreted as a replacement for echocardiography or clinician judgment. Its strongest value lies in preliminary screening, referral prioritization, remote monitoring, and community-based cardiovascular risk assessment. The paper concludes that AI-enabled wearable ECG screening may support earlier LVSD detection when integrated with validated algorithms, secure data systems, clinician oversight, and structured echocardiography referral pathways.