STATISTICAL METHODS FOR CLINICAL STUDIES

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): The identification and evaluation of interventions to reduce mortality and incidence of cancer is of critical public interest. Practical tools for addressing some new and continuing challenges in the design and analysis of clinical studies will be developed. A major emphasis will be placed on the design for Phase 1-111 clinical trials. The new developments will include strategies for early clinical studies of new biologic agents, designs for single arm survival studies, and solutions to several previously unresolved Phase III design issues. Flexible statistical design software will also be developed. New statistical methods for the joint analysis of longitudinal and time to event data in the context of Phase III studies will be investigated. The methods will take a Bayesian approach and utilize Markov Chain Monte Carlo (MCMC) sampling algorithms. There will be development and evaluation of exploratory survival analysis methods. New algorithms for constructing and interpreting prognostic subgroups of patients will be considered. Methodologies for model selection and for combining covariates in clinical association studies of moderate dimensions will also be investigated. Other topics proposed arise directly from our collaborative work on clinical trials. They will include analysis for time within a positive disease state and methods for non-parametric covariate adjustment. Collectively, the project will contribute to improvements in evaluating efficacy of cancer therapies though better methods for design, conduct and analysis of clinical studies.