Model Performances for Each Classification
| Classification | AUC | Sensitivity (%) | Specificity (%) | Accuracy (%) | PPV (%) | NPV (%) |
|---|---|---|---|---|---|---|
| IDH-positive vs. IDH-negative (LR_L2 and 5 features selected) | 0.831 (0.790–0.873) | 77.0 (72.5–81.4) | 70.7 (63.0–78.3) | 74.7 (71.4–78.4) | 82.5 (78.5–86.7) | 70.9 (64.9–76.6) |
| 1p19q-positive vs. 1p19q-negative (SVM_RBF and 15 features selected) | 0.724 (0.669–0.782) | 55.0 (43.0–66.0) | 72.3 (66.7–78.5) | 67.0 (62.5–71.2) | 39.6 (30.4–48.3) | 84.2 (80.5–88.2) |
PPV = positive predictive value; NPV = negative predictive value; IDH-positive = IDH mutant gliomas; IDH-negative = IDH wild-type gliomas; LR_L2 = logistic regression with L2 regularization algorithm; 1p19q-positive = 1p/19q codeleted gliomas; 1p19q-negative = gliomas not 1p/19q codeleted; SVM_RBF = support vector machine with radial basis function kernel algorithm.
Data are mean followed by 95% CI in parentheses.