Machine learning models for predicting acute kidney injury after pediatric living donor liver transplantation in biliary atresia

Figure 5 SHapley Additive exPlanations analysis explanation diagram of the best model. A: The ranking of the importance of different features for acute kidney injury prediction; B: The diagram of the influence direction and intensity of different features; C: The prediction analysis of a single sample. Mean SHapley Additive exPlanations (SHAP) value is the SHAP mean/feature importance, base value is the baseline value, higher indicates higher risk, Lower indicates lower risk. Cr: Creatinine; AST: Aspartate aminotransferase; LAC: Lactate; Ca²⁺: Calcium ion; FFP: Fresh frozen plasma; SHAP: SHapley Additive exPlanations.