Application Of Generalized Exponential Proportional Hazards Model In Partly Interval Censored Data Of Danish Diabetes

In survival analysis,researchers mainly focused on the patients with the onset of time or the time of death,but because of the limitation of time,money,or other reasons can not be accurately observed time of the incident,so there will be censored data.Censored data in the field of medicine is one of the common data types,the censored data usually leads to the loss of information of the research object,so it is very important to study this kind of data.In survival analysis of relevant research,to the censored data,the parameter and the nonparametric statistical inference has had certain development.However,the data type studied in my paper is partly interval censored data,which includes exact observation data and interval censored data,and there are relatively few studies on statistical inference of its parameters and non-parameters.In this paper,we apply this data type,consider the classical maximum likelihood estimation method,and calculate its parameters estimation value.This contents of article principally studies from the following three aspects:The first part considers partly interval censored data and studies the simulation effect of scale parameters under two conditions: one is that it is independent of covariates and the other is that it is correlated with covariates.In these two cases,the generalized exponential distribution model and generalized exponential scale parameter regression model are established,and the maximum likelihood method and Newton-Raphson algorithm are used to work out the estimation parameters iteratively.Then the model are applied to the Danish diabetes and AIDS case data sets,and the corresponding conclusions have great application value for the prevention of the two diseases.The second part considers partly interval censored data and applies Cox proportional hazard regression model,when the baseline hazard function is fixed,a parametric model is constructed;when the baseline hazard function is changes,a semi-parametric model is constructed.In this chapter,we considers the case when the baseline hazard function is fixed,the classical maximum likelihood method and Newton-Raphson algorithm are used to work out the estimated parameters iteratively,and the validity of the model method is proved by the simulation results.Finally,the model and estimation method are applied to the Danish diabetes data and AIDS data sets.The results of this example show that the model and research method have good effectiveness and robustness,and have practical significance.The third part considers partly interval censored data and applies a more flexible generalized accelerated hazard model.On particular occasion,this model can be converted into proportional hazard model,accelerated failure time model and accelerated hazard model.There are few relevant researches on this model itself,and most of them are statistical inference based on its particular model.In this model,the baseline hazard function takes into account the function relevant to the generalized exponential distribution,and applies the partial linearity to the model.Parameter deduction considers the combination of Newton-Raphson iteration and B-spline method.The four evaluation indexes and B-spline fitting results in the simulation results show that the statistical inference of this model has a good effect under this data type.Finally,the model and its method are applied to the AIDS case data set to consider its practical significance.