Skip to main content
. 2022 Jan 21;11:756828. doi: 10.3389/fonc.2021.756828

Table 4.

Summary of the subtype-specific classification performance of the combined model.

Molecular subtypes Cohorts AUC BAL_ACC SEN SPE
IDHwt Training 0.8414 (0.7906-0.8922) 0.7911 (0.6876-0.8476) 0.7895 (0.6184-0.9342) 0.7927 (0.6062-0.9119)
Testing 0.8623 (0.7453-0.9792) 0.8924 (0.8182-0.9697) 0.8889 (0.7222-1.0000) 0.8958 (0.7917-0.9792)
IDHmt-noncodel Training 0.8190 (0.7604-0.8776) 0.7761 (0.7212-0.8699) 0.6395 (0.5000-0.8488) 0.9126 (0.6632-0.9781)
Testing 0.8056 (0.6833-0.9278) 0.8066 (0.7424-0.9091) 0.7083 (0.5000-0.9167) 0.9048 (0.7381-1.0000)
IDHmt-codel Training 0.8193 (0.7695-0.8692) 0.7552 (0.6877-0.8104) 0.7944 (0.5514-0.9159) 0.7160 (0.5802-0.9198)
Testing 0.8036 (0.6898-0.9173) 0.8095 (0.7121-0.8939) 0.8333 (0.5833-0.9583) 0.7857 (0.6667-0.9524)

BAL_ACC, SEN, and SPE are short for balanced accuracy, sensitivity, and specificity, respectively. The 95% confidence interval for each index is shown.