The influence of uncertainties on optimization of vaccinations on a network of animal movements

In this article, multiobjective optimization of vaccinations is studied using graph-based modelling and simulations of the spreading of the disease. Real-life dataset of animal movements between farms and pastures in the Piedmont region of Italy is used, from which a dynamic network of contacts is reconstructed. Evolutionary multiobjective optimization of vaccinations is compared with vaccination strategies based on degrees or strengths of graph nodes, number of animals in the farms as well as with the ring vaccination strategy. In the article, the influence of uncertainties represented by the lack of knowledge of initial disease cases and the change of the contacts network by a rewiring process on the vaccination optimization is studied. Results of experiments show that evolutionary optimization of vaccinations can outperform vaccination strategies when enough information is provided. When many disease cases remain unknown or when the changes in the contacts network are large, the performance of the optimization algorithm is adversely affected. Obtained results motivate further research on modelling changes in animal movement patterns, as well as hybrid methods combining evolutionary optimization with vaccination strategies.