Translating Deep Learning to Clinical Practice: External Validation and Clinical Benefit of an Electrocardiogram-Based Neural Network for Detecting Low Ejection Fraction

Published: 23 Sept 2025, Last Modified: 18 Oct 2025TS4H NeurIPS 2025EveryoneRevisionsBibTeXCC BY 4.0
Keywords: ECG, electrocardiogram, cardiac screening, machine learning in healthcare, low ejection fraction detection, convolutional neural networks, ensemble learning, model generalizability, net benefit
TL;DR: We externally validate a deep learning model that detects low ejection fraction from ECGs and shows clear clinical benefit over standard screening methods.
Abstract: Low ejection fraction (EF), an indicator of impaired heart function, often goes undiagnosed and can lead to avoidable heart failure and arrhythmias. We developed and externally validated a deep learning model for detecting low EF from 12-lead electrocardiograms. The model achieved 85.8% sensitivity and 83.0% specificity on an independent validation cohort, with consistent results across demographic subgroups. These findings supported FDA 510(k) clearance of the model. Clinical net benefit analysis further showed that the model provides greater clinical value than default screening approaches, confirming its meaningful potential impact for clinical practice.
Submission Number: 11
Loading