Fig. 5

Identification of potential diagnostic biomarkers. A Ten-fold cross-validation of tuning parameter selection in the LASSO model. Each curve corresponds to one gene. B LASSO coefficient profile. The solid vertical line indicates the standard error of the partial likelihood deviation. The dashed vertical line is drawn at the optimal λ value. C, D SVM-RFE algorithm for validating biomarker gene expression. E RF plot for the relationship between the number of trees and the error rate. F Arrangement based on the relative importance of genes. G Venn diagram of intersected genes identified by LASSO, SVM-RFE, and RF algorithms. H, I Diagnostic efficacy of the ROC curves of diagnostic biomarkers in training set and validation set, respectively