Close

Agent-based modelling for COVID19 in South America

Abstract

The severity, complexity and the global reach of COVID-19 present unprecedented challenges to public health practitioners. As the pandemic continues and the world grapples with the spread of SARS-COV-2 and the catastrophic health and economic impacts of COVID-19, it has become apparent that a “silver bullet” solution is unrealistic. Human behavioral aspects which have been found to have a key role in the transmission and spread of respiratory virus, which are context dependent and subject to change over time. In this proposal, we focus on the transmission dynamics of COVID-19 in a metropolitan area in South America (Buenos Aires, Argentina) where mitigation strategies implemented in high-income countries might not be as successful or unsustainable for the period required to control transmission, due to the large informal economy and other cultural and political aspects. To address these issues and better understand the transmission dynamics in metro areas in South America, we have organized an interdisciplinary network of researchers and practitioners led by the Universidad Nacional de General Sarmiento, Argentina.

 

As part of this network, we developed an agent model of COVID-19 where each individual has contacts with other individuals inside a bipartite network of houses and works, in the tradition of the agent-based models proposed by Ferguson et al. The agents have states which are reminiscent of the compartments used in mean-field models. We used a modified version of the SEIR model proposed by Mordecai et al. Most of the parameters that are non-context specific such as the times of the individuals between each state and the proportions of asymptomatic and hospitalized were derived from the literature. For sensible parameters like the transmission rate and the number of hours spent on travel and at work, we fitted the model using surveillance data of the Buenos Aires metro area. Using this parameter set we can obtain scenarios showing different levels of exposure of individuals that represent different levels of quarantine. In the case of the Buenos Aires metro area, we can identify different stages based on the public health decisions made since March 2020. We also model the saturation of the health system with a threshold of the number of beds after which the mortality of those hospitalized increases radically. This allows us to assess the levels of quarantine (openness of the economy, mobility) that will keep the health system working properly. Up-to-date results of our model are available at https://covid19ungs.github.io/ariadnaNL.