Research Article
Automatic Detection of Atrial Fibrillation from ECG Signal Using Hybrid Deep Learning Techniques
Table 2
ResNet model accuracy and F1 score variation across five-fold cross-validation.
| Validation number | No. of epochs | Accuracy (%) | F1 score (%) |
| 1 | 5 | 66.39 | 75.61 | 10 | 69.62 | 15 | 76.60 | 20 | 81.23 |
| 2 | 5 | 60.70 | 77.87 | 10 | 68.04 | 15 | 75.60 | 20 | 81.17 |
| 3 | 5 | 62.64 | 80.23 | 10 | 69.43 | 15 | 77.81 | 20 | 82.80 |
| 4 | 5 | 68.21 | 83.54 | 10 | 72.42 | 15 | 78.03 | 20 | 83.11 |
| 5 | 5 | 64.52 | 85.67 | 10 | 77.77 | 15 | 82.84 | 20 | 84.40 |
| Overall F1 score (%) | 80.58 |
|
|