Fig. 4

Machine learning of prognostic gene signatures based on lactate metabolism. A Dynamic process of variable screening by Least Absolute Shrinkage and Selection Operator (LASSO) regression analysis. B, C The cross-validation results of LASSO regression (B) and coefficients of selected genes (C). D, E The OS curves of the lactate-related gene signatures in TCGA-LUAD samples (D) and GSE26939 samples (E). F Patients with lung adenocarcinoma were divided into high-risk and low-risk groups, taken the best cutoff value as the threshold. G The receiver operating characteristic (ROC) curves for 1-year, 2-year, 3-year and 5-year OS. H Scatter-plot showed survival time of patients in high-risk and low-risk groups. I Heatmap of the differential expressions of the 7 lactate-related genes selected to construct a prognostic gene signature in lung adenocarcinoma