Fig. 4

Construction of NS diagnostic models based on DEGs and IRGs. a Top 20 DEGs sorted by mean decrease accuracy based on random forest method. b Top 15 IRGs sorted by mean decrease accuracy based on random forest method. c Receiver operating characteristic (ROC) curves were calculated to evaluate the diagnostic efficiency of the DEG and IRG gene signatures with the training dataset. d AUC values of both models obtained by 1000 repeated tests based on bootstrap methods were shown in the density plot to validate the conclusions. AUC, area under the curve; CI, confidence interval; DEG, differentially expressed gene; IRG, immune-related gene