This research project will focus on the development of statistical techniques for assessing the effects of covariates and potential risk factors that are tested in complex longitudinal studies. The project has three distinct goals: 1. The adjustment of long term survival experiments for random effects such as common genetic or environmental factors which induce a dependency among some members of the population. Parametric and semi-parametric approaches based on the proportional hazards and an accelerated failure time model will be developed. 2. The development of a formalized modeling procedure for complex illness- recovery processes. This will lead to an extension of the causal modeling methods, such as path analysis, to survival analysis. 3. The development of improved techinques for comparing the survival experience of a prospective clinical study in which it is not feasible to randomize patients to conservative treatments or a historical control group. In each of these three areas, a study of both the large and small sample properties will be given and appropriate software will be developed to implement these methods.
Showing the most recent 10 out of 74 publications
