Fig. 9

Identification of immunotherapy−related signature genes. To discern genes relevant to immunotherapy, a composite of machine learning methods was employed to formulate a model predicting the efficacy of immunotherapy across seven distinct clinical immunotherapy cohorts. NaiveBayes was identified as the optimal algorithm, with the highest average AUC value of 0.651 (A). Thirteen insertion genes in the prognostic− and immunotherapy−related signatures were identified as key genes (B). The majority of key genes exhibited differential expression between normal and tumor samples sourced from the TCGA database (C−O). *p < 0.05; **p < 0.01; ***p < 0.001