Objective To explore prognostic factors of arteriovenous fistula(AVF)late failure in patients on maintenance hemodialysis based on random survival forest(RSF).Methods Select 156 patients on maintenance hemodialysis(MHD)who first received AVF in the First Affiliated Hospital of Guangxi University of Traditional Chinese Medicine from October 2018 to June 2022 as the research subjects.According to whether arteriovenous fistula late failure occurred,patients were divided into patency group(n=1 16)and failure group(n=40);explore the prognostic factors of AVF late failure.Results According to random survival forest results,the prognostic risk factors of arteriovenous fistula late failure were platelet count,triglyceride,fibrin,age,albumin,blood calcium,D-dimer and C reactive protein.The results of the Cox regression model showed that platelet counts(HR 1.02,95％CI 1.012-1.030,P＜0.001)and blood calcium(HR 8.84,95％CI 1.186-65.840,P=0.034)were independent influence factors of arteriovenous fistula late failure.According to the error prediction curve drawn by Brier score,the prediction performance of RSF was generally better than the Cox regression model.The effective prognostic factors determined by RSF-Cox model were platelet count,triglyceride,fibrin,age,albumin and blood calcium.Conclusion The random survival forest model has good selectability in the analysis of prognostic factors of arteriovenous fistula late failure.The random survival forest model established in this study can help clinicians identify patients with high-risk factors and carry out symptomatic intervention,which facilitates nursing and timely intervention to extend the use time of AVF.
random survival forestCox regression modelarteriovenous fistulalate failuremaintenance hemodialysis
国家科技期刊平台
NSTL
万方数据
