Skip to main content

Table 1 Performance evaluation results of trained models

From: Classification of COVID-19 electrocardiograms by using hexaxial feature mapping and deep learning

   Performance evaluation metrics
  Folds ACC PRE REC SPE F1-S AUC Loss MSE TT
Experiment 1 COVID-19 versus No-Findings via AlexNet Fold-1 91.00 88.68 94.00 88.00 91.26 97.92 0.283 0.090 103.91
Fold-2 95.00 92.45 98.00 92.00 95.15 96.44 0.680 0.050 101.50
Fold-3 93.00 89.09 98.00 88.00 93.33 97.16 0.451 0.070 101.56
Fold-4 94.00 94.00 94.00 94.00 94.00 97.30 0.533 0.060 101.47
Fold-5 95.00 94.12 96.00 94.00 95.05 98.60 0.317 0.050 101.44
Average 93.60 91.67 96.00 91.20 93.76 97.48 0.453 0.064 101.98
Experiment 2 COVID-19 versus No-Findings via proposed architecture Fold-1 94.00 90.74 98.00 90.00 94.23 97.36 0.721 0.060 105.10
Fold-2 96.00 92.59 100.0 92.00 96.15 99.60 0.203 0.040 103.41
Fold-3 97.00 96.08 98.00 96.00 97.03 99.84 0.151 0.030 102.83
Fold-4 98.00 98.00 98.00 98.00 98.00 99.88 0.086 0.020 102.86
Fold-5 96.00 94.23 98.00 94.00 96.08 99.08 0.300 0.040 101.87
Average 96.20 94.33 98.40 94.00 96.30 99.15 0.292 0.038 103.21
Experiment 3 COVID-19 versus No-Findings using only 2D ECG spectral images Fold-1 84.83 82.62 88.22 81.44 85.30 93.56 0.348 0.152 532.43
Fold-2 81.50 78.46 85.68 77.50 81.91 91.65 0.650 0.177 526.35
Fold-3 81.85 79.08 84.82 77.77 81.85 91.98 0.592 0.173 525.54
Fold-4 80.17 80.12 82.26 77.89 81.18 87.16 0.785 0.237 528.07
Fold-5 77.06 76.80 79.53 74.43 78.14 84.75 0.845 0.362 529.78
Average 81.08 79.42 84.10 77.81 81.68 89.82 0.644 0.220 528.43
Experiment 4 Negative versus Positive via proposed architecture Fold-1 92.00 88.89 96.00 88.00 92.31 94.08 0.831 0.080 106.95
Fold-2 95.00 92.45 98.00 92.00 95.15 98.04 0.297 0.050 103.28
Fold-3 91.00 88.68 94.00 88.00 91.26 95.00 0.520 0.090 103.43
Fold-4 94.00 90.74 98.00 90.00 94.23 94.06 0.663 0.060 103.71
Fold-5 93.00 92.16 94.00 92.00 93.07 93.70 0.827 0.070 102.23
Average 93.00 90.58 96.00 90.00 93.20 94.98 0.628 0.070 103.92
  1. Loss cross-entropy loss, TT training time (s), and all ACC, PRE, REC, SPE, F1-S, and ROC-AUC values are given as %
Back to article page

BMC Medical Informatics and Decision Making

ISSN: 1472-6947

Contact us