TABLE 4.
Summary of the subgroup-specific classification performance of the RLHC model.
| Molecular subgroups | Cohorts | Area under the curve | ACC (%) | SEN (%) | SPE (%) |
| Wingless | Training | 0.8277 (0.7133–0.9421) | 85.87 (53.26–93.48) | 66.67 (53.33–100.00) | 89.61 (44.16–97.40) |
| Testing | 0.9097 (0.7987–1.0000) | 80.00 (70.00–96.67) | 83.33 (83.33–100.00) | 79.17 (62.50–100.00) | |
| Sonic hedgehog | Training | 0.8997 (0.8236–0.9758) | 90.22 (65.22–95.65) | 75.00 (68.75–100.00) | 93.42 (57.89–98.68) |
| Testing | 0.8654 (0.6609–1) | 86.67 (56.67–100.00) | 75.00 (75.00–100.00) | 88.46 (50.00–100.00) | |
| Group 3 | Training | 0.8803 (0.8059–0.9547) | 81.52 (76.09–90.22) | 87.50 (65.00–97.50) | 76.92 (67.31–98.08) |
| Testing | 0.6652 (0.4667–0.8636) | 70.00 (60.00–83.33) | 100.00 (35.71–100.00) | 50.00 (25.00–100.00) | |
| Group 4 | Training | 0.9202 (0.8372–1.00) | 92.39 (82.61–96.74) | 76.19 (71.43–100.00) | 97.18 (80.28–100.00) |
| Testing | 0.6736 (0.3714–0.9759) | 86.67 (50.00–93.33) | 33.33 (16.67–100.00) | 100.00 (41.67–100.00) |
RLHC model means the model combining the 11 selected radiomics features, the tumor location, the hydrocephalus information, and the clinical factors. ACC, SEN, and SPE are short for accuracy, sensitivity and specificity, respectively. The 95% confidence interval for each index is shown.