Abstract: OBJECTIVE To analyze the risk factors for multidrug-resistant organisms (MDROs) infections in the perianal abscess patients and construct the machine learning prediction model so s to provide bases for clinical individualized treatment. METHODS The clinical data were collected from 540 patients with perianal abscess who received surgical procedures in Nanjing Traditional Chinese Medicine Hospital from Jan. 1, 2022 to Jun. 30, 2023 and were randomly divided into the training set and the test set in a 7∶3 ratio. The characteristic variables were screened out by Lasso algorithm, the problems of class imbalance were corrected by SMOTE algorithm, and nine types of machine learning prediction models were constructed. The performance of the prediction models was assessed by means of area under the curves (AUC), sensitivity, specificity, accuracy, F1 score, calibration curve and decision curve analysis (DCA), the output of the best model was interpreted by shapley additive explanations(SHAP). RESULTS Among the 540 patients with perianal abscess, 109 had MDROs infections. History of diabetes mellitus, extent of abscess, triglyceride and previous history of perianal abscess were the four major indexes for prediction of MDROs infections. Support vector machine (SVM) model showed the best performance among all the models(AUC=0.772,95%CI:0.675 to 0.868), with the prediction probability well-calibrated, and the DCA showed that it had high net benefit. SHAP analysis further revealed that the extent of abscess and history of diabetes mellitus had the greatest impact on the prediction of the models. CONCLUSION The SVM model is constructed and explained by SHAP, which may provide a quantitative tool for clinical identification of the perianal abscess patients with MDROs infections and optimization of use of antibiotics.