Construction of a machine learning-based predictive model for postoperative infections in elderly patients undergoing laparoscopic cholecystectomy
Objective To construct a machine learning-based predictive model for infections after laparoscopic cholecys-tectomy in elderly patients and identify the important variables affecting postoperative infection,providing insights for pharma-ceutical care.Methods This retrospective study included patients who underwent laparoscopic cholecystectomy in the Depart-ment of General Surgery of the First People's Hospital of Yancheng from December 2021 to January 2023.Patients were divid-ed into infected and non-infected groups according to their postoperative outcomes.SPSS Modeler 18.0 software was used to randomly divide the patients into a training set and a validation set in a 7∶3 ratio.Four machine learning models—random for-est,bayesian network,support vector machine,and classification and regression tree—were developed using the training set,and their performances were evaluated with the validation set.Results A total of 385 patients were finally included,including 62 patients in the infected group and 323 patients in the non-infected group.There were significant differences in the history of diabetes,acute gallbladder inflammation in perioperative period,preoperative white blood cell count,albumin level,total protein level,and operation duration between the infected group and the non-infected group(P<0.05).The random forest model,with an area under the receiver operating characteristic curve of 0.811,a specificity of 0.874,and an accuracy of 0.841,was selected as the final model.Variable importance was ranked as follows:preoperative albumin level,white blood cell count,history of diabetes,acute gallbladder inflammation in perioperative period,operation duration and preoperative to-tal protein levels.Conclusion The random forest model effectively predicts postoperative infections in elderly undergoing lapa-roscopic cholecystectomy.Clinical pharmacists should tailor pharmaceutical services based on changes in important variable characteristics.
laparoscopic cholecystectomypharmacy servicesmachine learningrandom forest modelrational medi-cation use
NETL
NSTL
万方数据
