Fig. 1

Using machine learning methods to analyze all available data. A Distribution of LASSO coefficients for variables. B Partial likelihood bias of the LASSO coefficient distribution. The vertical dashed line indicates the minimum partial likelihood deviation. C and D Random Forest for variable selection. Features identified by the Random Forest model according to the mean decrease in Gini index and Mean Decrease Accuracy for prediction of ICB outcomes. E Importance matrix plot of ICB’s predictors in XGBoost model among non-small lung cell cancer patients. NK, natural killer; NKT, natural killer T; CEA, carcinoembryonic antigen; CA125, carbohydrate antigen 125; CA199, carbohydrate antigen 199; NSE, neuron-specific enolase; LASSO, least absolute shrinkage and selection operator; XGBoost, eXtreme Gradient Boosting, RF, random forest