Fig. 6From: Development and validation of ‘Patient Optimizer’ (POP) algorithms for predicting surgical risk with machine learningFeature importance for any complication: XGBoost gain (left) and SHAP (right), where each dot represents one sample, the colour indicates the value and the position on the x-axis indicates the impact (positive or negative) on model output. Refer to Table 4 for terminologyBack to article page