Fig. 2
The normalized values of feature importance analysis based on (A) Random Forest, (B) Decision Tree, (C) K-Nearest Neighbor, and (D) XGBoost models in surgical site infection outcome. Transfusion: Transfusion in hospitalization. Intra-operative…: Intra-operative complication. The performance metrics of the machine learning models are as following: Random forest: Accuracy: 0.71, Sensitivity: 0.81, Specificity: 0.37; Decision tree: Accuracy: 0.58, Sensitivity: 0.64, Specificity: 0.37; K-nearest kneighbour: Accuracy: 0.58, Sensitivity: 0.64, Specificity: 0.37; XG Boost: Accuracy: 0.64, Sensitivity: 0.65, Specificity: 0.70.
