Table 1 Prediction results on the fivefold test sets for models trained with both view and location contrast.
AUROC | AUPRC | Sensitivity | Specificity | Precision | F1 Score | MCC | |
|---|---|---|---|---|---|---|---|
Results on the fivefold test sets [mean (± standard deviation)] | |||||||
Image group | 0.820 | 0.342 | 0.747 | 0.748 | 0.154 | 0.201 | 0.195 |
(± 0.182) | (± 0.173) | (± 0.339) | (± 0.097) | (± 0.103) | (± 0.142) | (± 0.171) | |
Patient | 0.819 | 0.586 | 0.883 | 0.716 | 0.339 | 0.489 | 0.431 |
(± 0.119) | (± 0.211) | (± 0.145) | (± 0.075) | (± 0.058) | (± 0.079) | (± 0.131) | |
Out-of-fold results on the whole dataset | |||||||
Image group | 0.654 | 0.148 | 0.302 | 0.767 | 0.123 | 0.175 | 0.048 |
Patient | 0.807 | 0.424 | 0.875 | 0.717 | 0.333 | 0.483 | 0.426 |
