Abstract 13330: Deep Learning Algorithm Predicts Angiographic Diameter Stenosis in Stable Patients Using Only a Standard 12-lead Electrocardiogram

Abstract

Introduction: Current standard electrocardiogram analysis algorithms cannot predict the presence and extent of coronary artery disease (CAD), especially in stable patients. Hypothesis: A novel artificial intelligence algorithm (ECGio) is able to predict the presence, location, and severity of coronary artery lesions in a minimally selected stable patient population as verified by coronary angiography. Methods: A cohort of 1659 stable outpatients were randomly divided into training (86%) and validation (14%) subsets, maintaining population characteristics. ECGio was trained and then validated using electrocardiograms paired with angiograms collected from electronic medical records. Coronary artery lesions were evaluated by a continuous measure of stenosis ranging from 0% - 100%. The prediction was then compared to the angiogram result (worst diameter stenosis in each vessel) with the error calculated per patient and per vessel. Results: In the primary analysis of the validation cohort, 22 had no angiographic CAD and were grouped with 56 patients with mild CAD (DS ≤30%), 31 had moderate CAD (DS 30-70%), and 113 had severe CAD (DS ≥70%). On a vessel level ECGio was able to predict stenosis severity with an overall average error of 16% as well as vessel specific errors of 18% in the LAD, 19% in the LCX, 18% in the RCA, and 8% in the Left Main. Absolute Error and Left Main error can be seen in the Figure. Conclusion: This validation study strongly suggests that it is possible to utilize an artificial intelligence algorithm to determine the presence and severity of CAD in stable patients using data from a standard 12- lead electrocardiogram.