Statistical Inference Of Two Kinds Of Censored Data Based On Transformation Model

In survival analysis,medical workers and biostatistics' experts are most concerned about the patient's survival time.With the rapid development of medicine industry and science or technology,on the premise of ensuring the patient's long survival time,medical workers and biostatistics' experts began to pay attention to medical expense,so that medical workers would select the optimal treatment according to a comprehensive program of survival time and medical costs.On the other hand,a variety of censored data exists in survival analysis,there are interval censored and doubly censored data except known the right censored data.Doubly censored data are common in many fields,especially the progress of the disease research and epidemiological studies.Therefore,this paper will give the relevant research of censored medical costs and doubly censored data.The first part of this paper introduces the characteristics and research status of data lifetime medical costs;then a semiparametric transformation model is given based on the medical costs;then the error term in the likelihood function of observed data structure subject to normal distribution,we can use MH sampling algorithm to sample from posterior distribution about the parameters based on Bayesian estimation theory.Estimation of the regression parameter value and conversion fitting image function could be obtained by simulation of sampling,and simulation results show the effectiveness of the estimation method.Finally,the method is applied to the data of medical cost of the Multicenter Automatic Defibrillator Implantation Trial(MADIT)in the United States,and the practicability of the model is proved.The second part of this article firstly introduces the doubly censored data and its current situation,then gives the quantile regression model of doubly censored data,establishes the quantile regression model between doubly censored data with Box-Cox transformation and covariate,constructs the target function for solving quantile parameters,and estimates transformed parameter and regression coefficients respectively by two steps of optimization.A large number of simulation studies have verified the effectiveness of the proposed model method.