Agent-based model and data assimilation: Analysis of COVID-19 in Tokyo

TL;DR

This paper presents an agent-based model combined with a particle filter for analyzing COVID-19 spread in Tokyo, introducing a new method to evaluate the effective reproduction number and other unknown parameters.

Contribution

It introduces a novel data assimilation approach using particle filters with an agent-based model to study COVID-19 dynamics and estimate key epidemiological parameters.

Findings

Effective reproduction number can be accurately estimated.

Unknown parameters and populations are effectively evaluated.

Model stability and uncertain quantities are thoroughly analyzed.

Abstract

In this paper we introduce an agent-based model together with a particle filter approach for studying the spread of COVID-19. Investigations are performed on the metropolis of Tokyo, but other cities, regions or countries could have been equally chosen. A novel method for evaluating the effective reproduction number is one of the main outcome of our approach. Other unknown parameters and unknown populations are also evaluated. Uncertain quantities, as for example the ratio of symptomatic / asymptomatic agents, are tested and discussed, and the stability of our computations is examined. Detailed explanations are provided for the model and for the assimilation process.