Understanding the Role of Daily Activities in the Transmission of COVID-19 in Urban Settings Using an Agent-Based Model

The COVID-19 pandemic has caused widespread disruption and prompted the implementation of nonpharmaceutical measures in several countries to contain the spread of the disease until a vaccine became available. Urban mobility reduction has historically been employed to limit the transmission of epidemics. However, only some studies have quantified the effectiveness of such restrictions across different sectors of society. To address this gap, we have adapted an agent-based model that utilizes data from the Google Community Mobility Report to simulate the dynamics of COVID-19 across 14 Brazilian capitals. Each agent in the model has a network of contacts established from mobility data across various categories. We simulated six scenarios for each capital, each with different probabilities of contagion in each category. Our findings show that different scenarios are more effective in describing the curve of infected people and deaths in each city. In particular, our results indicate that the same scenario was optimal for describing the number of cases and deaths in Belo Horizonte. However, the first peak in the number of deaths could not be reproduced in the model, possibly due to issues with the data recording. Our proposed model can be further developed to incorporate additional elements related to the dynamics of an epidemic and can serve as an additional tool in understanding and planning actions to contain the spread of COVID-19 beyond mobility, particularly in urban centers.