Value of establishing a warning model for developing urosepsis after renal endoluminal surgery for renal stones based on random forest algorithm
Objective To investigate the value of establishing a warning model for developing urosepsis after endoluminal surgery for renal stones based on random forest algorithm.Methods A total of 362 patients with renal stone who underwent endoluminal surgery were selected as research subjects.According to whether the patients developed urosepsis after surgery,they were divided into urosepsis and non-urosepsis groups.Clinical data of the patients were collected.Multivariate logistic regression was used to screen for risk factors affecting postoperative urosepsis.R software was used to establish a random forest model for predicting postoperative urosepsis in patients who underwent renal endoluminal surgery,and the efficacy of the model was verified.Results There were 34 patients who developed urosepsis after renal endoluminal surgery,with an incidence rate of 9.39％.The differences in age,gender,diabetes,stone diameter,number of stones,infectious stones and operation time were significant different between urosepsis and non-urosepsis groups(P＜0.05).Multivariate logistic regression results showed that age,gender,diabetes,stone diameter,number of stones,infectious stones and operation time were independent risk factors for urosepsis in patients with renal stones after renal endoluminal surgery(P＜0.05).The random forest model revealed that the order of the importance in predicting urosepsis after renal endoluminal surgery was the operation time,diabetes,infectious calculi,gender,stone diameter,number of stones and age.Both models had good predictive performance.Conclusion The risk prediction model established by the random forest algorithm has good predictive performance for urosepsis after endoluminal surgery.
random forestrenal stonesendoluminal surgeryurosepsismultivariateearly warning modelpredictive efficacy
NETL
NSTL
万方数据
