The Construction Of Prognosis Prediction Model Of Idiopathic Membranous Nephropathy And The Effect Of Traditional Chinese Medicine Decoction Intervention

Objective: 1.To investigate the prognostic factors of idiopathic membranous nephropathy(IMN);2.To explore the influence of TCM decoction intervention on the prognosis of IMN patients;3.The predictive ability of COX proportional hazard regression model and random survival forest(RSF)model was compared.Methods: Retrospective analysis of 101 patients in our hospital who diagnosed with IMN and Compliance with inclusion and exclusion criteria From January 2018 to December 2021.Clinical data of IMN patients(including demographic data,clinical indicators and pathological data of patients during renal puncture biopsy and treatment methods during follow-up)were collected and pretreated.COX proportional hazard regression model and random survival forest model were constructed to analyze the data,and the factors affecting the prognosis of IMN were discussed.The KM survival curve of high and low risk groups in the prediction model was drawn to compare the survival between the two groups.The KM curve of treatment methods was drawn to compare the survival of different treatment groups.The predictive performance of the two models was compared using the timedependent subject operation curve(td ROC)area under the curve(AUC).Results: In this study,a total of 101 patients with IMN who Compliance with inclusion and exclusion criteria were included.By the end of observation time,35 patients(34.7%)had renal function progression,66 patients(65.3%)had stable renal function.There were 30 cases(29.7%)in the intervention group of traditional Chinese medicine decoction and 71 cases(70.3%)in the non-traditional Chinese medicine decoction group.The results of COX proportional hazard regression model and random survival forest model suggested that high cystatin C concentration,low serum albumin level and advanced age at renal biopsy were independent risk factors for renal function progression.It was also confirmed that the intervention of TCM decoction in the treatment regimen for IMN patients was a protective factor for renal function progression.COX proportional hazard regression model showed that the risk of end-point events increased 1.031 times for each year of age increased at the time of renal biopsy.Patients with albumin <20g/L at renal biopsy had a 2.738-fold increased risk of endpoint events compared with IMN patients with albumin≥20g/L.The risk of endpoint events increased by 6.063 times for every 1mg/L increase in cystatin C concentration during renal biopsy.The risk of endpoint events in IMN patients without TCM decoction intervention was 2.812 times higher than that in IMN patients with TCM decoction intervention.In the random survival forest model,the four most important screened variables were cystatin C concentration(VIMP =0.037),albumin <20g/L(VIMP =0.033),TCM decoction intervention(VIMP =0.029),and age(VIMP =0.011).The KM curves of the high-low risk group of the two models were drawn,both of which showed that the models could significantly distinguish the survival rates of patients in the high-risk group and the lowrisk group,with statistical significance,and the high-risk group was significantly correlated with poor prognosis.The KM curve of the treatment groups showed that the TCM decoction intervention group had more survival advantages.td ROC curves and AUC values of the two models were drawn,showing that the AUC values of COX proportional hazard regression model in the first 1-3 years were 0.860,0.877 and 0.821,respectively,and their AUC values in the first 1-3 years were all above 0.8,indicating a high predictive performance of the models.The AUC values of the random survival forest model in the first 1-3 years were 0.928,0.908 and 0.897,respectively,and the AUC values under the td ROC curve in the first 1-3 years were all above 0.89,indicating that the prediction performance of the model was high.The AUC values of the two prediction models were compared,and the prediction accuracy and stability of the random survival forest model were better than that of the COX proportional hazard regression model.Conclusions:1.Age,serum albumin level,cystatin C concentration and treatment regimen are independent predictors of renal function progression in IMN patients.2.Advanced age,low serum albumin level,and high cystatin C concentration increase the risk of renal function progression in patients.Intervention with Chinese herbal decoction is a protective factor for prognosis in patients with IMN,and has survival advantages.3.The prediction accuracy and stability of random survival forest model are better than those of COX proportional risk regression model.