Table 3.
Predictive performance of the longitudinal fusion radiomic models and clinical-radiomics models based on the early phase.
| AUC | Sensitivity | Specificity | Accuracy | |
|---|---|---|---|---|
| Longitudinal fusion radiomic-models (training/validation) | ||||
| Pre + delta | 0.866/0.750 | 0.929/0.833 | 0.623/0.700 | 0.723/0.738 |
| 1st + delta | 0.903/0.839 | 0.821/0.667 | 0.871/0.933 | 0.857/0.857 |
| Pre + 1st + delta | 0.871/0.869 | 0.750/0.667 | 0.871/0.900 | 0.837/0.833 |
| Clinical-radiomic model (training/validation) | ||||
| Pre + clinical | 0.762/0.569 | 0.679/0.333 | 0.757/0.900 | 0.735/0.738 |
| 1st + clinical | 0.801/0.767 | 0.786/0.833 | 0.771/0.633 | 0.776/0.691 |
| Delta + clinical | 0.934/0.864 | 0.927/0.750 | 0.829/0.867 | 0.857/0.833 |
| Pre + delta + clinical | 0.870/0.736 | 0.821/0.667 | 0.757/0.800 | 0.776/0.762 |
| 1st + delta + clinical | 0.908/0.828 | 0.821/0.667 | 0.900/0.967 | 0.878/0.881 |
| Pre + 1st + delta + clinical | 0.866/0.864 | 0.714/0.667 | 0.886/0.900 | 0.837/0.833 |