The spread of an infectious disease such as COVID-19 is governed by complex social interactions that are challenging to model. Policy makers must take measures to control the spread of infection despite the unknowns that accompany a novel epidemic.

Agent-based models capture the intricacies of social interaction for a subset of the population by defining the behavior of individual agents. While they can be computationally expensive for large populations, their outcomes are stochastic. Therefore, they can be used to test disease prevention policies, that can be difficult to simulate using deterministic approaches. We developed an agent-based model that is inspired by several interactive simulations on the internet used as pedagogical tools for describing the COVID-19 pandemic. We define metrics to estimate the socio-economic cost of disease prevention policies on the population. We present a policy-making tool based on blackbox optimization that provides well-rounded intervention measures in terms of socio-economic cost and disease control.

Several intervention measures are suggested by the algorithm with varying degrees of disease control and socio-economic cost. Policy makers can choose an intervention measure based on their preference. This research recommends the use of mathematical search algorithms for identifying the critical amount of intervention necessary to control infectious diseases and formulate intervention policies that minimize socio-economic cost.