Table 2.
Ablation study for highlighting the importance of ECOC and bootstrapping in HCN framework
| Fusion strategy | Normal (%) | Bacterial pneumonia (%) | Viral pneumonia (%) |
|---|---|---|---|
| Sensitivity | |||
| HCN-DML w/o ECOC | 95.05 | 93.20 | 96.88 |
| HCN-DML w/o Bootstrapping | 95.10 | 96.04 | 97.94 |
| HCN-DML w/o ECOC + Bootstrapping | 94.06 | 92.23 | 95.83 |
| HCN-DML | 96.08 | 95.10 | 98.96 |
| HCN-FM w/o ECOC | 92.08 | 87.74 | 97.85 |
| HCN-FM w/o Boot strapping | 94.06 | 90.38 | 97.89 |
| HCN-FM w/o ECOC + Bootstrapping | 97.00 | 92.00 | 83.00 |
| HCN-FM | 91.51 | 95.05 | 100.00 |
| Precision | |||
| HCN-DML w/o ECOC | 96.00 | 96.00 | 93.00 |
| HCN-DML w/o Bootstrapping | 97.00 | 97.00 | 95.00 |
| HCN-DML w/o ECOC + Bootstrapping | 95.00 | 95.00 | 92.00 |
| HCN-DML | 98.00 | 97.00 | 95.00 |
| HCN-FM w/o ECOC | 93.00 | 93.00 | 91.00 |
| HCN-FM w/o Boot strapping | 95.00 | 94.00 | 93.00 |
| HCN-FM w/o ECOC + Bootstrapping | 88.20 | 86.80 | 98.80 |
| HCN-FM | 97.00 | 96.00 | 93.00 |