Table 4.
Model performance in EGFR mutation prediction using OOB data
| AUC (95% CI)a | P for difference in AUC | |
|---|---|---|
| All cases (n = 478) | ||
| Intratumor | 0.730 (0.682–0.777) | |
| Intra+peritumor | 0.774 (0730–0.817) | < 0.001 ( vs intratumor) |
| Intra+peritumor+clinical | 0.826 (0.788–0.864) | 0.005 ( vs clinical) |
| Clinical | 0.777 (0.734–0.821) | |
| Adenocarcinoma only (n = 348) | ||
| Intratumor | 0.630 (0.572–0.689) | |
| Intra+peritumor | 0.687 (0.632–0.743) | < 0.001 ( vs intratumor) |
| Intra+peritumor+clinical | 0.744 (0.693–0.796) | 0.045 ( vs clinical) |
| Clinical | 0.703 (0.648–0.758) |
AUC, area under the curve; CI, confidence interval; EGFR, epidermal growth factor receptor; OOB, out-of-bag.
a
AUC was calculated from OOB data, namely a cohort that was not used in building the model.