Table 3 Results of RF classifier construction model.
Group | Model | acc | bac | f1 | spe | pre | recall | mse | auc |
|---|---|---|---|---|---|---|---|---|---|
Training set | Clinical | 0.7274 | 0.6297 | 0.4511 | 0.8746 | 0.6706 | 0.3847 | 0.5216 | 0.7645 |
Radiomics | 0.9323 | 0.9243 | 0.8846 | 0.9456 | 0.9160 | 0.9030 | 0.2573 | 0.9633 | |
Combined models | 0.9247 | 0.9088 | 0.8673 | 0.9476 | 0.9102 | 0.8701 | 0.2733 | 0.9726 | |
Validation set | Clinical | 0.7735 | 0.6540 | 0.4857 | 0.9217 | 0.7254 | 0.3863 | 0.4758 | 0.7358 |
Radiomics | 0.9119 | 0.9040 | 0.8478 | 0.9217 | 0.8837 | 0.8863 | 0.2967 | 0.9558 | |
Combined models | 0.9245 | 0.9197 | 0.7696 | 0.9304 | 0.8986 | 0.9090 | 0.2747 | 0.9694 |