Training and testing accuracy and classification tree modeling of cytokines. (A) Feature importance for discrimination between sepsis (N = 80) and CRS (N = 16) cohorts. Cytokines are listed in terms of decreasing feature importance (right axis), and the performance of the model is shown (training and testing accuracy) when predictors are added to the model 1 at a time (left axis). (B) Classification tree model for discrimination between sepsis (N = 80) and CRS (N = 16) cohorts.