Statistical Inference Of Several Semiparametric Models Under Recurrent Event Data With A Terminal Event

Recurrent event data,such as tumor recurrence,recurrent asthma attacks,and multiple hospital visits in clinical medicine,refer to a specific type of complex data in survival analysis,which are frequently encountered in biomedicine,economics,social science and other related fields.In order to analyze such complex data,it is essential to develop appropriate statistical models based on the structure and background of the data and sensible statistical inference procedures,aiming to achieve the goals of explanation and prediction.Therefore,investigating such complex data has important theoretical significance and broad applications.In this thesis,we propose some new modeling and inference methods for recurrent event data with a terminal event.In Chapter 2,we propose a class of semiparametric mixed rate models to evaluate the overall effects of covariates on the weighted composite endpoint for multiple types of recurrent and terminal events.The mixed model is flexible since it allows the covariate effects on the rate function of the weighted composite process to be proportional or convergent while leaving the dependence structure among the recurrent and terminal events unspecified.For inference on the model parameters,the unbiased estimating equations are developed and the asymptotic properties of the resulting estimators are established.The finite sample performance of the proposed procedure is evaluated through number simulation studies.We apply the proposed method to areal data set on a medical cost study of chronic heart failure patients for illustration.In Chapter 3,we use a class of semiparametric transformation models with time-varying coefficients to analyze the weighted composite endpoint for recurrent and terminal events.In this model,we let the effects of covariates on weighted composite event process be time-varying.We model the mean function of the composite event and then construct unbiased estimating equations to estimate the regression coefficients.Additionally,we derive the asymptotic properties of the resulting estimators.Finally,the empirical performance of the proposed method is verified through numerical simulations,followed by an application to a set of actual data from a bladder cancer study.In Chapter 4,we discuss the analysis of recurrent event process in the present of multiple intermittent gaps and a terminal event.A flexible semiparametric additive-multiplicative model is assumed for the rate function of recurrent events.Based on the idea of generalized estimating equation,we establish unbiased estimating equations for the estimation,which do not need any assumptions about the correlation between recurrent and terminal events.The large sample properties of the estimators are developed.We assess the finite-sample performance of the proposed method through simulation studies.An application to a medical cost study of chronic heart failure patients is also provided.