Table 2 Model performance: ECG-WMA-Net versus qualitative and quantitative ECG metrics

From: Identification of cardiac wall motion abnormalities in diverse populations by deep learning of the electrocardiogram

 

ECG-WMA-Net

ECG qualitative

ECG quantitative

AUC

0.781 (0.762, 0.799)

0.571 (0.552, 0.590)

0.681 (0.658, 0.705)

Accuracy

0.756 (0.747, 0.764)

0.808 (0.799, 0.817)

0.622 (0.611, 0.632)

F1 Score

0.362 (0.343, 0.381)

0.228 (0.203, 0.253)

0.269 (0.251, 0.287)

Sensitivity

0.652 (0.624, 0.680)

0.271 (0.240, 0.301)

0.665 (0.632, 0.698)

Specificity

0.768 (0.759, 0.777)

0.871 (0.863, 0.879)

0.617 (0.605, 0.628)

NPV

0.949 (0.944, 0.954)

0.911 (0.904, 0.918)

0.940 (0.933, 0.947)

PPV

0.250 (0.235, 0.267)

0.197 (0.174, 0.221)

0.168 (0.155, 0.182)

  1. AUC area under the curve, ECG electrocardiogram, NPV negative predictive value, PPV positive predictive value, WMA wall motion abnormality
Back to article page