Fig. 1

Generation and validation of the mutation-based gene set. A An overview of the propensity score algorithm used to balance confounding factors, including age, cancer types, and drug types. B Survival analysis of the mutation-based gene set in the training cohort. C Survival analysis of the mutation-based gene set in different age groups (age < 65 and age ≥ 65). D Survival analysis of the mutation-based gene set in different drug type subgroups (anti-PD-1, anti-CTLA-4 and combination). E Survival analysis of the mutation-based gene set in different cancer type subgroups (urothelial cancer, melanoma, non-small-cell lung cancer, renal cell carcinoma, and colorectal cancer). F Survival analysis of the mutation-based gene set in the validation cohort. G The proportion of patients who responded to ICI therapy in the high-risk and low-risk groups in the training cohort. H The distribution of risk scores in groups with different ICI clinical response statuses in the training cohort. I The proportions of patients with a response to ICI therapy in the high-risk and low-risk groups in the validation cohort. J The distribution of risk scores in groups with different ICI clinical response statuses in the validation cohort