Key result
A multi-input deep learning model using 12-lead ECG data detected significant aortic regurgitation with an AUC of 0.802 (95% CI 0.762-0.837), significantly outperforming a 2D-CNN model (P<0.001).
Why the study?
Aortic regurgitation burden may increase with population aging, creating a need for an effective screening method using electrocardiography.
Does a multi-input deep learning model using 12-lead ECG data improve the detection of significant aortic regurgitation compared to a 2D-CNN model alone?
Population
29,859 paired ECG and echocardiography records, including 412 AR cases
Comparison
Multi-input neural network model vs 2D-CNN alone and other machine learning models
Design
Machine learning model development and validation study
Authors
Loading...
May support ECG-AI screening for significant AR in large cohorts; leaves open prospective validation before clinical use.
Cross-Sectional (n=29,859)
Does a multi-input deep learning model using 12-lead ECG data improve the detection of significant aortic regurgitation compared to a 2D-CNN model alone?
Effect estimate: AUC 0.802 (95% CI 0.762-0.837)
Absolute Event Rate: 0.802% vs 0.734%
p-value: p=<0.001
A multi-input deep learning model using 12-lead ECG data can detect significant aortic regurgitation with modest predictive value, outperforming a standard 2D-CNN model.
Sawano et al. (2021) conducted a cross-sectional in Aortic regurgitation (n=29,859). Multi-input deep learning model (2D-CNN and FC-DNN) vs. 2D-CNN model alone and other machine learning models was evaluated on Detection of significant aortic regurgitation (AUC) (AUC 0.802, 95% CI 0.762-0.837, p=<0.001). A multi-input deep learning model using 12-lead ECG data detected significant aortic regurgitation with an AUC of 0.802 (95% CI 0.762-0.837), significantly outperforming a 2D-CNN model (P<0.001).
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: