Table 7 Per-class recognition performance comparison with other models.
From: A hybrid model combining 1D-CNN and BERT for intelligent ECG arrhythmia classification
Methods | Metric | N | S | F | V | Q |
|---|---|---|---|---|---|---|
CNN+Bi-LSTM30 | Precision | 99.60 | 82.80 | 87.50 | 94.57 | 98.54 |
Recall | 99.01 | 92.70 | 83.33 | 99.10 | 99.75 | |
F1-score | 99.30 | 87.40 | 85.36 | 96.78 | 99.14 | |
ECGTransform17 | Precision | 99.36 | 91.67 | 91.30 | 95.74 | 99.30 |
Recall | 99.46 | 86.91 | 82.68 | 97.65 | 99.06 | |
F1-score | 99.41 | 89.22 | 86.78 | 96.69 | 99.18 | |
MSH-GCN25 | Precision | 100.00 | 99.28 | 100.00 | 99.72 | 99.75 |
Recall | 99.98 | 99.28 | 100.00 | 100.00 | 99.75 | |
F1-score | 99.88 | 91.76 | 100.00 | 98.38 | 97.14 | |
CNN-BERT (ours) | Precision | 99.56 | 93.41 | 98.79 | 83.95 | 98.87 |
Recall | 99.75 | 92.28 | 97.01 | 85.01 | 99.42 | |
F1-score | 99.66 | 92.84 | 97.83 | 84.47 | 99.79 |
