Optimal Crossover Design for Generalized Linear Model with Binary Response

Jankar, Jeevan (2019) Optimal Crossover Design for Generalized Linear Model with Binary Response. Masters thesis, Indian Institute of Science Education and Research Kolkata.

PDF (MS dissertation of JEEVAN JANKAR (14MS060)) 14MS060.pdf - Submitted Version Restricted to Repository staff only Download (987kB)

Abstract

In this thesis, we identify locally optimal crossover designs for generalized linear models. Generalized linear models are widely used for modeling the responses that are binary in nature. Pharmaceutical industries frequently conduct clinical trials where the outcome is either success or failure of a particular therapy. In this thesis, generalized linear models are used for such crossover experiments with binary responses. We use generalized estimating equations to estimate the model parameters along with their variances. Different correlation structures have been proposed to capture real situations.We identify optimal allocation of patients for different sequences of treatments. For two-treatment crossover designs, it has been shown via simulations that the optimal allocations are reasonably robust to the choices of the correlation structures. We discuss a real example of a multiple treatment crossover experiment using Latin square designs. A simulation study was presented to show that the optimization technique developed here is more efficient when responses from the same subject are more correlated.

Actions (login required)

View Item