Abstract: Objective To establish and validate a model based on machine learning for predicting the risk of septic shock in patients with multidrug-resistant Klebsiella pneumoniae (MDR-KP) infection. Methods A total of 1 385 patients with MDR-KP infection hospitalized at the Affiliated Huaian No.1 People's Hospital of Nanjing Medical University fromJanuary 2019 to April 2024 were retro-spectively enrolled in the study. The key predictive factors were identified using the least absolute shrinkage and selection operator (LASSO) regression and Random Forest Recursive Feature Elimination(RF-RFE). Then,the multiple machine learning models,including logistic regression(LR)，decision trees(DT)，random forests (RF)，extreme gradient boosting（XGBoost)，support vector machines (SVM)，K-nearest neighbors（KNN)，and light gradient boosting machines (LightGBM)，were constructed. The performances of these models were evaluated using the five-fold cross-validation method. The area under the receiver operating characteristics (ROC) curve（AUC)，accuracy，sensitivity，specificity,positive predictive value,negative predictive value,and Fl score of these models were also evaluated. ResultsSix key predictive factors, including difficulty in weaning from mechanical ventilation, use of vasoactivedrugs，coagulation disorders,acute kidney injury，septicemia,and the level of procalcitonin,were identified.Therandomforest model performed the best in the validation set, with an AUC value of O.864. ConclusionA model for predicting the risk of septic shock in patients with MDR-KP infection is successfully constructed, providing an effective tool for early identification of high-risk patansa more influencing factors to improve its generalizability.