Mitchell Aaron Schepps, UCLA

• Title: Metaheuristics for finding efficient longitudinal designs for bipolar patients with and without a genetic covariate treated with sustained release lithium

• Abstract: When there are a few candidate designs for implementation in pharmacometrics, a common method to select the design is to adopt a model-based approach and determine the design with the best value of a pre-selected design criterion among the candidate designs. The design criterion is formulated as a scalar function the Fisher information matrix, which can be challenging to evaluate for non-linear mixed effects models. We propose using nature-inspired metaheuristic algorithms to search for efficient model-based designs with user selected number of time points to optimize the design criterion. We discuss use of metaheuristics as a general purpose optimization tool and apply it to design efficient longitudinal studies for bipolar patients with and without a genetic covariate and treated with lithium.