Fig. 12

Development of a prognostic risk model based on EXOSC family co-expression genes in HNSC using TCGA cohort data. (*p < 0.05, **p < 0.01, ***p < 0.001). (A) Univariate Cox regression analysis of the 12 prognostic EXOSC family co-expression genes, presented as forest plots of hazard ratios (HRs) with 95% confidence intervals. (B) Feature coefficients across varying lambda values in LASSO regression. (C) Partial likelihood deviance versus log(λ) in the LASSO Cox regression model. (D) Risk stratification analysis as follows: Top: Patient distribution into high-risk (red) and low-risk (blue) groups based on median risk score. Middle: Survival status (alive = blue; deceased = red). Bottom: Expression heatmap of the 5 prognostic genes across risk groups. (E) Kaplan-Meier analysis demonstrating significant survival difference between risk groups (log-rank test). (F) Time-dependent receiver operating characteristic (ROC) curves assessing model predictive performance.