Multiobjective Optimization of a Targeted Vaccination Scheme in the Presence of Non-diagnosed Cases

摘要

One of the problems in disease prevention and epidemics control is the large level of uncertainty present in the solved problems. If an optimization of targeted vaccination scheme is attempted, the evaluation of solutions usually involves simulating the non-deterministic spreading of the disease. However, the non-determinism of the epidemic dynamics is just one of the sources of uncertainty in this problem. This paper studies another source of uncertainty in optimization of targeted vaccination schemes, which is the existence of non-diagnosed cases of the disease. In the paper the multiobjective optimization of a targeted vaccination scheme is carried out under an assumption that only a fraction of really existing cases of the disease are known. Solutions found by the optimization algorithm in this limited-knowledge scenario are then evaluated in a simulation in which all existing cases of the disease are taken into account. Optimized solutions are compared with the mass vaccination scheme which is not based on the knowledge of existing cases of the disease. An uncertainty-handling technique is proposed that randomly adds infected individuals to the simulations in order to compensate for the unknown ones.