Fig. 4

Identification of nine immune infiltration-related prognostic genes to build a risk score model. A Forest plot of prognostic module genes with P < 0.01 by univariate Cox regression. B, C Ten-time cross-validation of the LASSO model and coefficient profile of filtered prognostic genes. D Construction of the prognostic model by multivariate Cox regression. The hazard ratios (HRs) and 95% confidence intervals (CIs) of each candidate gene are shown. E The multivariate cox regression of the association between clinical factors (including the risk score) and survival. F The optimal cut point selected by the maximum standard log-rank statistics in HNSCC cohort. G PCA based on 277 immune infiltration-related genes showing different immune phenotypes in two risk groups. H The model gene expression heatmap combined with the distribution of risk scores and the survival of patients in two risk groups. I Kaplan-Meier survival curves of two risk groups in the whole and stage-divided HNSCC cohorts. J ROC curves based on risk score in HNSCC cohort within 1–5 years. K Nomogram combining risk score with clinical information. L ROC curves evaluating the predictive efficacy of the nomogram for the overall survival within 1–5 years