A multi-method approach to modeling COVID-19 disease dynamics in the United States

Abstract In this paper, we proposed a multi-method modeling approach to community-level spreading of COVID-19 disease. Our methodology was composed of interconnected age-stratified system dynamics models in an agent-based modeling framework that allowed for a granular examination of the scale and se...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Amir Mokhtari, Cameron Mineo, Jeffrey Kriseman, Pedro Kremer, Lauren Neal, John Larson
Formato: article
Lenguaje:EN
Publicado: Nature Portfolio 2021
Materias:
R
Q
Acceso en línea:https://doaj.org/article/508d5bc1527646498fbe55f99bde001e
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
Descripción
Sumario:Abstract In this paper, we proposed a multi-method modeling approach to community-level spreading of COVID-19 disease. Our methodology was composed of interconnected age-stratified system dynamics models in an agent-based modeling framework that allowed for a granular examination of the scale and severity of disease spread, including metrics such as infection cases, deaths, hospitalizations, and ICU usage. Model parameters were calibrated using an optimization technique with an objective function to minimize error associated with the cumulative cases of COVID-19 during a training period between March 15 and October 31, 2020. We outlined several case studies to demonstrate the model’s state- and local-level projection capabilities. We further demonstrated how model outcomes could be used to evaluate perceived levels of COVID-19 risk across different localities using a multi-criteria decision analysis framework. The model’s two, three, and four week out-of-sample projection errors varied on a state-by-state basis, and generally increased as the out-of-sample projection period was extended. Additionally, the prediction error in the state-level projections was generally due to an underestimation of cases and an overestimation of deaths. The proposed modeling approach can be used as a virtual laboratory to investigate a wide range of what-if scenarios and easily adapted to future high-consequence public health threats.