Research Article
Automatic Detection of Atrial Fibrillation from ECG Signal Using Hybrid Deep Learning Techniques
Table 3
Performance of hybrid model (ResNet and bidirectional LSTM) using five-fold cross-validation.
| Validation number | Epochs | Accuracy (%) | F1 score (%) |
| 1 | 5 | 59.14 | 71.73 | 10 | 66.79 | 15 | 72.33 | 20 | 79.26 |
| 2 | 5 | 62.44 | 77.97 | 10 | 69.36 | 15 | 73.57 | 20 | 80.02 |
| 3 | 5 | 61.47 | 79.85 | 10 | 70.61 | 15 | 76.27 | 20 | 81.11 |
| 4 | 5 | 64.56 | 84.93 | 10 | 71.24 | 15 | 78.69 | 20 | 82.33 |
| 5 | 5 | 61.29 | 85.94 | 10 | 69.55 | 15 | 76.34 | 20 | 82.87 |
| Overall F1 score (%) | 80.08 |
|
|