Extended Data Table 6 Comparing the machine learning model’s performance in predicting LV systolic dysfunction (EF < 50%) for different feature categories

From: Synthetic generation of cardiac tissue motion from surface electrocardiograms

  1. We report the sensitivity, specificity, F1 score, accuracy, and AUC with 95% CI, reduction in echocardiography ((True Negatives + False Negatives) / total samples), and missed rate (False Negatives / total samples) in the detection of LV systolic dysfunction (positive class = presence of LV systolic dysfunction). AUC = area under the receiver operator characteristic curves, CI = confidence interval, ECG = electrocardiogram, TDI = tissue Doppler imaging.
Back to article page