Particle swarm based algorithms forfinding locally and Bayesian D-optimal designs
Published Web Locationhttps://doi.org/10.1186/s40488-019-0092-4
When a model-based approach is appropriate, an optimal design can guide how tocollect data judiciously for making reliable inference at minimal cost. However, findingoptimal designs for a statistical model with several possibly interacting factors can beboth theoretically and computationally challenging, and this issue is rarely discussed inthe literature. We propose nature-inspired metaheuristic algorithms, like particle swarmoptimization (PSO) and its variants, to solve such optimization problems. Wedemonstrate that such techniques, which are easy to implement, can find differenttypes of optimal designs for models with several factors efficiently. To facilitate use ofsuch algorithms, we provide computer codes to generate tailor made optimal designsand evaluate efficiencies of competing designs. As applications, we apply PSO and findBayesian optimal designs for Exponential models useful in HIV studies and re-design acar-refuelling study for a Logistic model with ten factors and some interacting factors.