Table 2.
Model performance of different modules included in the proposed framework.
| Benchmark | Attention | DS Aggregator | Accuracy (%) |
Sensitivity (%) |
Specificity (%) |
NPV (%) |
PPV (%) |
AUC |
|---|---|---|---|---|---|---|---|---|
| ✓ | 67.14 | 73.89 | 60.30 | 67.38 | 67.77 | 0.6710 | ||
| ✓ | ✓ | 69.64 | 72.22 | 65.93 | 70.45 | 69.80 | 0.7023 | |
| ✓ | ✓ | 75.71 | 69.29 | 81.18 | 72.93 | 83.09 | 0.7336 | |
| ✓ | ✓ | ✓ | 76.43 | 71.41 | 80.36 | 74.16 | 81.74 | 0.7422 |
Values in bold black font represent the best performance in each column.