A pan-cancer computational model for predicting immunotherapy response by quantifying tumor immune activity

• Pan-cancer computational model “Immune Thermometer” quantifies tumour immune activity by integrating immune infiltration and stromal suppression features. • This model generates continuous digital scores for individual tumours and enables immune activity to be measured independently of cancer type. • This model accurately identifies immune-active tumours, outperforming existing biomarkers such as PD-L1 and TMB, and predicts diverse immunotherapy responses. • This model represents a clinically actionable tool for patient stratification and provides a unified foundation for systematic investigations of immune biology.