Statistical Problems in Multivariate Survival Analysis

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): This project will develop and investigate new methodology for analyzing recurrent event data from cardiovascular disease, asthma study and other biomedical research. The proposal describes four projects. The first project concerns statistical inferences for multiple-type recurrent events data. Parameter estimation will be based on an estimating equations approach. To increase efficiency, weighted estimating equations will be developed with weights inversely proportional to intra-subject covariance; parametric and non-parametric correlation estimators will be developed. Inference will be based on the multivariate central limit theorem and modern empirical processes theory. Asymptotic and finite sample properties will be examined. The proposed methods will be used to analyze data from a clinical study of left ventricular dysfunction patients and a retrospective cohort study of childhood asthma. The second project considers an accelerated failure time marginal means model and conditional multiplicative means model for analyzing censored recurrent event data which allow for terminating events. Parameter estimation will be based on an estimating equation approach. The strengths and weaknesses of the proposed method will be critically examined via theoretical investigations and simulation studies. Data from a study of kidney transplant patients will be analyzed using the proposed methods. The third project concerns marginal and conditional means models for analyzing censored recurrent event data, which accommodate both terminating events and dependent censoring. Parameter estimation will be conducted through an estimating equation approach, with inference based on the multivariate central limit theorem and empirical processes. Asymptotic and finite sample properties will be examined. Methods proposed will be applied to analyze data from a clinical study of dialysis patients. The fourth project investigates additive means models for censored recurrent event data. We will propose methods of estimation, which are applicable for recurrent events with independent censoring only, with terminal events and independent censoring, and with both terminal events and dependent censoring. Asymptotic and finite sample properties will be examined. Data sets from asthma, dialysis and renal transplant studies will be analyzed using the proposed methods.