The Lead Episode 35: A Novel ECG-Based Deep Learning Algorithm to Predict Cardiomyopathy in Patients With Premature Ventricular Complexes-JACC Clinical Electrophysiology

JACC Clinical Electrophysiology

Pdf Summary

Researchers have developed a deep learning algorithm that can predict cardiomyopathy in patients with premature ventricular complexes (PVCs) using just a 12-lead electrocardiogram (ECG). PVCs are early depolarizations of the ventricular myocardium and are prevalent in 1% to 4% of the general adult population. PVC-induced cardiomyopathy (PVC-CM) is a condition in which patients with idiopathic cardiomyopathy see improvements in left ventricular ejection fraction (LVEF) with suppression of PVCs. The algorithm was trained and tested using electronic medical records from 5 hospitals, with internal training and testing performed at one hospital and external validation performed at the others. The dataset included a total of 14,241 patients with documented PVCs. The primary outcome of the study was the first diagnosis of LVEF reduction to less than 40% within 6 months. The deep learning model achieved an area under the receiver operating curve of 0.79 for predicting LVEF reduction to less than 40%. The model performance was consistent across different demographic groups, including sex and race. The algorithm's prediction was independent of PVC burden, highlighting the importance of the QRS complex and ST-segment during sinus rhythm rather than PVC morphology. The researchers also conducted a manual review of patients who underwent PVC ablation and found that the majority (60%) experienced an improvement in LVEF post-ablation. The deep learning algorithm shows promise for accurately predicting cardiomyopathy in patients with PVCs and could potentially aid in clinical decision-making for these patients. However, further studies are needed to validate and assess the clinical impact of this algorithm.

Keywords

deep learning algorithm

cardiomyopathy

premature ventricular complexes

PVCs

12-lead electrocardiogram

PVC-induced cardiomyopathy

left ventricular ejection fraction

electronic medical records

QRS complex

ST-segment