Comparison of CART decision tree model and logistic regression model for predicting frailty risk in hospitalized patients with cirrhosis
Objective To construct fateful risk prediction models for inpatients with cirrhosis by CART decision tree and logistic regression respectively,and compare the prediction effects of 2 models.Methods A total of 317 pa-tients with cirrhosis who met the admission criteria in a tertiary A hospital from March to October 2023 were select-ed by convenience sampling method,and CART decision tree and logistic regression were used to construct a frailty risk prediction model for inpatients with cirrhosis.In internal verification,the method of Bootstrap resampling 1 000 times was used to compare the performance of 2 models by using accuracy,sensitivity,specificity,positive predic-tive value,negative predictive value,and area under the curve(AUC)of receiver operating characteristic(ROC).Result The accuracy of CART decision tree and logistic regression were 86.3％and 89.1％,the sensitivity was 76.5％and 92.5％,the specificity was 91.8％and 82.8％,the positive predictive value was 84.8％and 91.6％,the negative predictive value was 86.8％and 83.8％,and the AUC was 0.876 and 0.965.Conclusion logistic model is better than CART decision tree model in predicting frailty risk in inpatients with cirrhosis,which can provide ref-erence for early screening and prevention of frailty.
regression analysisdecision treeliver cirrhosis diseaseenervationprediction model
万方数据
维普
