Table 5 Performance comparison of different attention-based models on Train and Independent datasets across various evaluation metrics.
Dataset | Model | AUC (%) | ACC (%) | MCC (%) | Sensitivity (%) | Specificity (%) | Precision (%) | F1 (%) |
|---|---|---|---|---|---|---|---|---|
Train Dataset | AttXGBoost | 86.62 | 82.00 | 64.20 | 76.15 | 87.40 | 83.84 | 79.81 |
AttDenseFusion | 91.65 | 83.00 | 67.82 | 92.66 | 74.80 | 75.94 | 83.47 | |
AttGRU | 93.58 | 86.00 | 71.13 | 86.24 | 85.04 | 83.19 | 84.68 | |
AttMLP | 91.99 | 83.05 | 65.90 | 81.65 | 84.25 | 81.65 | 81.65 | |
AttBiLSTM | 91.35 | 85.59 | 71.46 | 88.99 | 82.68 | 81.51 | 85.09 | |
AttBiLSTM_DE(LR-guided Optimization) | 94.74 | 86.86 | 74.25 | 91.74 | 82.68 | 81.97 | 86.58 | |
AttBiLSTM_DE(AttBiLSTM-guided Optimization) | 94.40 | 86.91 | 73.85 | 91.60 | 82.60 | 81.91 | 86.50 | |
Independent Dataset | AttXGBoost | 97.28 | 92.00 | 77.25 | 70.13 | 98.85 | 94.74 | 80.60 |
AttDenseFusion | 97.63 | 94.66 | 84.45 | 81.82 | 98.46 | 94.03 | 87.50 | |
AttGRU | 95.19 | 90.00 | 75.00 | 95.00 | 88.00 | 70.19 | 81.00 | |
AttMLP | 97.99 | 94.36 | 84.08 | 88.31 | 96.15 | 87.18 | 87.74 | |
AttBiLSTM | 98.04 | 95.55 | 87.12 | 83.12 | 99.23 | 96.97 | 89.51 | |
AttBiLSTM_DE(LR-guided Optimization) | 98.48 | 95.85 | 88.00 | 87.01 | 98.46 | 94.37 | 90.54 | |
AttBiLSTM_DE(AttBiLSTM-guided Optimization) | 98.39 | 96.44 | 87.75 | 86.89 | 98.55 | 92.98 | 89.83 |
