Head and neck squamous cell carcinoma (HNSCC) exhibits a poor 5-year survival rate. TNFRSF4 is gaining attention in tumor therapy. The objective of this study was to forecast the expression of TNFRSF4 in HNSCC tissue using analysis of pathological images and investigate its possible molecular mechanisms. Transcriptome, clinical, and pathological data of HNSCC patients from the TCGA database were analyzed. Features were extracted with PyRadiomics for support vector machine model development. The evaluation of model performance was conducted using ROC curve, calibration curve, and decision curve analyses. The correlation between pathomics score (PS), patient prognosis, and immune- related genes was assessed. TNFRSF4 expression was significantly higher in the tumor group and indepen-dently associated with HNSCC prognosis. Features were extracted to build a predictive model for TNFRSF4, which demonstrated strong performance. PS correlated positively with immune-related genes. This research highlights the potential of TNFRSF4 as a prognostic factor and demonstrates the utility of PS in relation to immune-related genes.
Keywords: HNSCC; TNFRSF4; pathomics; prognosis; machine learning models.