Simulation of COVID-19 Spread Scenarios in the Republic of Kazakhstan Based on Regularization of the Agent-Based Model

We propose an algorithm for modeling scenarios newly diagnosed cases of COVID-19 in the Republic Kazakhstan. The is based on treating incomplete epidemiological data and solving inverse problem reconstructing parameters agent-based model (ABM) using set available data. main tool constructing ABM Covasim open library. In event a drastic change situation (appearance new strain, removal or introduction restrictive measures, etc.), are updated taking into account additional information previous month (online assimilation). solved by stochastic global optimization (of tree-structured Parzen estimators). As example, we give two propagation calculated December 12, 2021 period up to January 20, 2022. scenario that took New Year holidays (published http://covid19-modeling.ru ) almost coincided with what happened reality (the error was 0.2%).