Modeling future spread of infections via mobile geolocation data and population dynamics.An application to COVID-19 in Brazil.

Pedro S Peixoto,Sérgio M Oliva,Cláudia Peixoto,Diego Marcondes,Haroldo V. Ribeiro

doi:10.1371/journal.pone.0235732

Abstract

Mobile geolocation data is a valuable asset in the assessment of movement patterns of a population. Once a highly contagious disease takes place in a location the movement patterns aid in predicting the potential spatial spreading of the disease, hence mobile data becomes a crucial tool to epidemic models. In this work, based on millions of anonymized mobile visits data in Brazil, we investigate the most probable spreading patterns of the COVID-19 within states of Brazil. The study is intended to help public administrators in action plans and resources allocation, whilst studying how mobile geolocation data may be employed as a measure of population mobility during an epidemic. This study focuses on the states of São Paulo and Rio de Janeiro during the period of March 2020, when the disease first started to spread in these states. Metapopulation models for the disease spread were simulated in order to evaluate the risk of infection of each city within the states, by ranking them according to the time the disease will take to infect each city. We observed that, although the high-risk regions are those closer to the capital cities, where the outbreak has started, there are also cities in the countryside with great risk. The mathematical framework developed in this paper is quite general and may be applied to locations around the world to evaluate the risk of infection by diseases, in special the COVID-19, when geolocation data is available.

Highlights

The COVID-19, caused by the coronavirus SARS-CoV-2, has spread quickly after its first reported cases in Wuhan, China, in December 2019, posing a serious threat to health systems and the world economy [1]
In this work we focus on extracting the mobility pattern between cities directly from mobile mobility data, so we are neither assuming node degree clustering to determine its dynamic behavior, nor assuming a mean field or statistical distribution to model the dynamics
The In Loco company provided anonymized data containing the geolocation of millions of users of their software development kit (SDK), which is present in many popular mobile apps

Summary

Introduction

The COVID-19, caused by the coronavirus SARS-CoV-2, has spread quickly after its first reported cases in Wuhan, China, in December 2019, posing a serious threat to health systems and the world economy [1]. It is a common sense that the pandemic should be fought in two frontiers: by saving lives while avoiding the collapse of health systems, and by protecting the population from the economic impacts of the pandemic, specially its most vulnerable parcel [2]. For either goal to be achieved, health officials and government authorities should have reliable information about. Modeling future spread of infections via mobile geolocation data and population dynamics under open-source license, available at https:// github.com/pedrospeixoto/mdyn. The mobile geolocation movement matrices are available within the same GitHub repository. Supporting information is provided at www.ime.usp.br/~ pedrosp/covid-19

Methods

Results

Conclusion