This project will develop and investigate new methodology for analyzing data from clinical studies of cancer and other chromic diseases. During this grant period methods for analyzing clustered failure time data will be emphasized. The specific areas of concentration are as follows: 1. Methods for the analysis of association structured in clustered failure time data. Clustered data might arise, for example, if there are institutional effects in multicenter clinical trials. Methods for non- parametric estimation of the association structure, methods for using these estimates to examine the association structure and help guide formulation of association models, and methods for local estimation of association parameters over time, will be investigated.. Estimates of association structure will also be applied to the problems of estimating weight matrices and variances of estimated parameters in transformed linear marginal regression analyses, and possibly other models as well. 2. Methods for semiparametric analysis of transformed linear failure time models with clustered data. Numerical issues, appropriate choice of weight functions, and methods for inference for rank based estimating equations will be addressed.
Showing the most recent 10 out of 15 publications
