A Flexible Agent-Based Model to Study COVID-19 Outbreak -- A Generic Approach

TL;DR

This paper presents a flexible agent-based model to simulate COVID-19 outbreak dynamics, allowing for analysis of various control measures and population factors to inform effective intervention strategies.

Contribution

It introduces a customizable agent-based simulation framework that incorporates diverse population and intervention parameters for studying COVID-19 spread.

Findings

Without lockdown, cases peak around week 6 and decline by week 15.

Lockdowns slow the infection peak and extend the outbreak duration.

Higher asymptomatic proportion reduces observed case numbers.

Abstract

Understanding dynamics of an outbreak like that of COVID-19 is important in designing effective control measures. This study aims to develop an agent based model that compares changes in infection progression by manipulating different parameters in a synthetic population. Model input includes population characteristics like age, sex, working status etc. of each individual and other factors influencing disease dynamics. Depending on number of epicentres of infection, location of primary cases, sensitivity, proportion of asymptomatic and frequency or duration of lockdown, our simulator tracks every individual and hence infection progression through community over time. In a closed community of 10000 people, it is seen that without any lockdown, number of cases peak around 6th week and wanes off around 15th week. If primary case is located inside dense population cluster like slums, cases peak early and wane off slowly. With introduction of lockdown, cases peak at slower rate. If sensitivity of identifying infection decreases, cases and deaths increase. Number of cases declines with increase in proportion of asymptomatic cases. The model is robust and provides reproducible estimates with realistic parameter values. It also guides in identifying measures to control outbreak in a community. It is flexible in accommodating different parameters like infectivity period, yield of testing, socio-economic strata, daily travel, awareness level, population density, social distancing, lockdown etc. and can be tailored to study other infections with similar transmission pattern.