A0266
Title: Discriminating among several random effects models
Authors: Chiara Tommasi - University of Milan (Italy) [presenting]
Abstract: Random effects models are largely applied across all disciplines, particularly in clinical studies and biosciences. The focus is on the design issue of optimally discriminating among several random effects models using the Kullback-Leibler divergence (KL) criterion. A theoretical result proves that the KL criterion leads to classical $T$ optimality based on a different inner product. A closed-form expression for the minimum Kullback-Leibler divergence is provided, which makes much easier the implementation of the algorithms to find out optimal designs. Finally, two examples show how to apply the proposed methodology. The first application concerns discrimination among fractional polynomials with a single continuous variable; the latter identifies the best design to discriminate among several multi-factorial random effects models.
