Título: KL-optimality for discriminating among random effect models
Authors: Jesus Lopez-Fidalgo
Co-authors: Victor Casero-Alonso, Sergio Pozuelo-Campos, Chiara Tommasi, and Weng Kee Wong
Language: en

The bulk of the work in optimal designs in the literature is focused on finding model based optimal designs for mixed effects models. These models include an implicit correlation structure of the observations. This work assumes there are several plausible models and we want to use the Kullback-Leibler (KL) divergence criterion to find a design that optimally discriminates among the competing models. The optimization problem is complex involving a multi-level nested optimization problem. To this end, we develop theoretical results that simplify the computational burden and implement a nature-inspired metaheuristic algorithm. We provide two applications for fractional polynomial models with a single continuous variable, and the second considers multi-factor random ffects models.