Artificial intelligence-assisted diagnosis and prognostication in low ejection fraction using electrocardiograms in inpatient department: a pragmatic randomized controlled trial

Comprehensive Summary

This study investigated the use of an artificial intelligence driven 12-lead electrocardiogram (ECG) in a cohort of patients at an academic medical center in Taiwan. The AI-ECG algorithm sought to identify patients at risk of heart failure, as determined by a left ventricular ejection fraction below 50%. In a randomized controlled trial, 13,631 patients were randomly assigned to an intervention group (n=6840) receiving input from the AI-ECG and a control group (n=6791) receiving standard care. The rate of detection of low ejection fraction (EF) was significantly increased in the intervention group, with the benefit most pronounced among patients categorized as high-risk by the AI algorithm. The rate of newly diagnosed low EF was 13.0% in the intervention group and 8.9% in the control group. The AI-ECG algorithm improved diagnostic efficiency and was especially valuable in populations at risk of missed or delayed diagnosis.

Outcomes and Implications

Heart failure affects over 6 million Americans and leads to approximately 1 million hospitalizations annually. Low levels of left ventricular ejection fraction is often a sign of heart failure as blood is not being pumped efficiently. Often, reduced ejection fraction is asymptomatic, so it is frequently underdiagnosed in its early stages, with asymptomatic ventricular dysfunction affecting 3–6% of the general population. This study offers a promising approach to AI-driven ECG interpretation, and can improve rates of early detection and better management of low EF.