Table 3.
Results of clinical model, radiomics model and DLRN predictive performance.
| Set | Model | AUC (95%CI) | ACC | ER | SEN | SPE | PPV | NPV | P |
|---|---|---|---|---|---|---|---|---|---|
| Training | DLRN | 0.936 (0.874-0.999) | 0.914 | 0.086 | 0.650 | 0.969 | 0.813 | 0.930 | Reference |
| Radiomics model | 0.914 (0.876-0.953) | 0.755 | 0.245 | 0.463 | 1.000 | 1.000 | 0.691 | 0.551 | |
| Clinical model | 0.696 (0.564-0.827) | 0.828 | 0.172 | 0.000 | 1.000 | NA | 0.828 | <0.001 | |
| External validation | DLRN | 0.833 (0.732-0.933) | 0.897 | 0.103 | 0.474 | 0.972 | 0.750 | 0.912 | Reference |
| Radiomics model | 0.799 (0.675-0.922) | 0.881 | 0.119 | 0.263 | 0.991 | 0.833 | 0.883 | 0.394 | |
| Clinical model | 0.664 (0.523-0.805) | 0.849 | 0.151 | 0.000 | 1.000 | NA | 0.849 | 0.034 |
CI, confidence interval; ACC, accuracy; ER, error rate; SEN, sensitivity; SPE, specificity; PPV, positive predictive value; NPV, negative predictive value; NA, not available.