Fig. 2

Details of radiomic feature extraction using LASSO (a) and XGBoost (b) at baseline. Two feature selection steps were applied to the extracted radiomic features with the least absolute shrinkage and selection operator (LASSO) and XGBoost. a The LASSO model is a linear combination of the selected features weighted by their respective coefficients. The x-axis denotes LASSO coefficients. Features with nonzero coefficients denote greater contributions to the model and were selected. b Feature importance evaluates how valuable each feature is in the construction of the gradient boosted decision trees within the XGBoost model and is calculated by information gain. The x-axis measures the information gain