Measurement error problems in modeling multivariate failure time data

点击翻译按钮获取中文摘要

We propose a research program on the measurement error problems in statistical analysis of multivariate failure time data. We will develop approaches to correct biases of parameter estimates induced by measurement error in covariates when using a marginal hazards model to analyze (correlated) multivariate failure time data. We will develop approaches to handle problems of missing covariates but with auxiliary information. We will conduct research work on missing value problems of generalized linear mixed models when auxiliary information to the missing values is available. We will investigate the asymptotic properties of the parameter esitmates. We will conduct simulation studies on small sample properties of them. The methods will be use to analyze real data for illustration. The outcome of our proposed research will provide a more efficient method to analyze data from medical and other studies. It can help medical and other investigators to better understand their underlying research program, that could ultimately lead to the design of better medical treatments and more efficient medicine. Thus, benefiting people of Canada and people of other countries.