Incorporating Mobility Information into an Epidemiological Model to Model Infected Cases at the Onset of COVID-19 Pandemic

Abstract

In the wake of the rampage of the COVID-19 pandemic, its ripple effects have changed every aspect of our lives. Restricting mobility becomes one of the primary countermeasures adopted by governments to mitigate the spreading of this airborne COVID-19 virus. Considering the undeniable benefits of restricting mobility, restricting mobility is still not a viable option for many countries in the world because of its heavy toll on economy and people’s mental health. This study proposes two models – Mobility SEIR (Susceptible – Exposed – Infected – Removed) and DLSTM-MSEIR (Deep Long Short-Term Mobility SEIR) based on the classical epidemiological models – SEIR to model the COVID-19 cases at the onset of the pandemic and reveal the importance of considering mobility information in the traditional and classical epidemiological model-SEIR. The modeling results show the two proposed models could outperform the classic SEIR model in modeling and predicting the COVID-19 cases at the onset of the pandemic and bring meaningful insights into reducing the spread of the pandemic. Moreover, the hybrid model DLSTM-MSEIR has a better performance on modeling COVID-19 cases than the Mobility SEIR model when the change point indications are less clear at the period of study.