Figure 2.

Variable importance for the eXtreme Gradient Boosting (XGBoost) model. Permutation variable importance for the most accurate model (XGBoost) scaled to a maximum of 100, which shows that age, number of ICU beds, creatinine, and lactate were the most important variables for predicting 28-d mortality when using data from the first 2 d of ICU admission. ALT = alanine transaminase, AST = aspartate aminotransferase, BMI = body mass index, CPK = creatine phosphokinase, CRP= c-reactive protein, P/F ratio = Pao₂/Fio₂ ratio.