An Agent-Based Model for Localized COVID-19 Transmission Dynamics and Intervention Impacts

Throughout the COVID-19 pandemic, disease-modeling has guided government health officials in choosing appropriate interventions.However, most current models simulate disease spread on a more generalized scale, lacking specificity for localities such as towns or counties, leading to one-size-fits-all policies being instituted on the country or state-wide-level.However, localities differ in many social determinants of health, which impact disease dynamics therefore necessitating models tailored to individual locations.This research aims to answer this question: what local factors affect COVID-19 outbreak severity and intervention effectiveness?To do this, a novel agent-based disease model was created using NetLogo to simulate contextualized COVID-19 disease dynamics at the local level.Model inputs include population demographic composition, area size, vaccination ratio, interventions (mask, test-and-isolate, or lockdown), and compliance rate.Agents representing the simulated local population are assigned specified traits, and become "susceptible", "exposed", "infected", "recovered", "quarantined", or "dead" as they interact with other agents.The model was validated using data from state and local health agencies for Westchester County, NY (84.2% accuracy).A sensitivity analysis demonstrated that a higher elderly population, a lower young population, a lower vaccination rate, and weaker interventions were all factors that increased outbreak severity.A comparison of selective localities representing metric axes of high/low age and high/low vaccination was conducted for four different U.S. counties and showed that 1) any intervention would dramatically reduce locality variations and 2) interventions have higher impact in higher risk localities.This model enables local officials to better focus limited resources when making health related decisions, and a website (www.localcovidmodel.org)has been created for model access.