Multiple Imputation For Censored Data

In the analysis of cure rate,there are many cases where the observation time is missing due to some reasons.For this situation,it is difficult to know whether the censored subjects are cured or uncured.That is to say,the individual’s susceptible indicator is missing data,so it is difficult to analyze the cure rate of this part of the data.Aiming at this kind of missing situation,a multiple imputation method is used.Survival data with missing data is used to fit cure rate quantile regression model to estimate the parameters in the cure rate quantile regression model.Firstly,it uses an iterative algorithm to estimate the conditional uncured probability of each individual,and then uses the estimated probability to estimate the individual’s susceptible indicator.That is to say,it classifies each subject as cured or uncured by using the conditional uncured probability and the Bernoulli sample imputation.Secondly,according to the data after imputation,the sample is divided into cured and uncured according to susceptible indicator.The parameters of the quantile regression model are estimated for the uncured individuals by using the locally weighted method.Finally,the imputation procedure is repeated multiple times.Then taking an average over the resultant estimators,it can obtain a consistent estimate of the cure rate quantile regression model with censored data.This method relaxes the common global linearity assumption,so quantile regression can be applied to any particular quantile of interest.In order to analyze this multiple filling method,this paper analyzed the estimated results based on simulation studies and case studies.This paper constructed a data set that satisfies some certain assumptions and used the data set to fit the model and obtained the parameter estimation results of the model.The results were compared with the set hypothesis values,and the effectiveness of the estimation method was determined by comparative analysis consistency.In addition to simulation studies,the model and estimation method were also applied to the rdata data set included in R and the data set of follicular cell lymphoma respectively.The model construction and parameter estimation were achieved by R and the estimation results were obtained,in order to analyze the multiple filling method of the cure rate quantile regression model.