A multiomics model for predicting immunotherapy response in advanced lung cancer

This study aimed to develop and validate a multi-omics integration-based model for predicting immunotherapy response in patients with advanced lung cancer. A total of 342 consecutive patients receiving immunotherapy were enrolled and randomly assigned to training and validation sets. Baseline clinical, inflammatory, and multi-omics parameters were collected. Univariate analysis followed by multivariate logistic regression identified independent predictors. Three machine learning models-random forest, support vector machine, and logistic regression-were constructed. Model performance was assessed using area under the receiver operating characteristic curve (AUC), calibration curves, and decision curve analysis (DCA). SHAP analysis interpreted feature contributions. Univariate analysis revealed that TMB, TIME score, NLR, PLR, LDH, and albumin significantly differed between responders and non‑responders ( P < 0.05). Multivariate logistic regression confirmed all six as independent predictors ( P < 0.05). Among the three models, random forest achieved the best performance, with AUC values of 0.835 (95% CI: 0.776–0.895) in the training set and 0.832 (95% CI: 0.740–0.925) in the validation set, outperforming logistic regression (0.822 and 0.757) and SVM (0.773 and 0.749). SHAP analysis ranked feature importance as NLR, TIME score, PLR, TMB, LDH, and albumin. Calibration curves showed good agreement between predicted and observed probabilities. DCA confirmed high clinical net benefit across a range of threshold probabilities. A multi-omics integration-based predictive model for immunotherapy response in advanced lung cancer was successfully developed and validated. The model exhibits satisfactory predictive performance and clinical applicability, serving as an objective, quantitative decision-support tool for individualized immunotherapy management.