Mining Mobility Data in Response to COVID-19

Introduction

As the new coronavirus spreading from coast to coast in the United States, over thirty million people have been confirmed a year after the COVID-19 outbreak. The federal government announced a national state of emergency in the middle of March followed by a major disaster in the New York metropolitan area. A month later, the COVID-19 pandemic has been determined as a disaster in all states throughout the nation, announced by the Federal Emergency Management Agency (FEMA, 2020). Following an exponential growth in the number of infected cases with an increasing fatality rate, as of the end of April, state governments have been taking action to minimize the transmission rate. Almost all the state executives have issued the stay-at-home or shelter-in-place order by mid-April 2020 (Mervosh et al., 2020). The purpose of issuing travel restrictions or lockdown policies is to mitigate the spread of COVID-19 by reducing the human mobility concentrated on decreasing the probability of the contact. The impact of such policies on the disease control of COVID-19 has been explored by using epidemic models, such as the SIR model. Although some studies suggest that lockdowns or travel restrictions have different and limited effects across regions with significant opportunity costs (Bonardi et al., 2020; Chinazzi et al., 2020), such policies can reduce the daily growth rate of infected cases in the United States (Courtemanche et al., 2020). Similar suggestions have been provided based on the simulation in Chapter 2, which indicate that the lockdown policy can flat the curve effectively without new medicines or vaccines. The purpose of the lockdown policy is to reduce the reproduction rate that can be reflected by the effectiveness of the restricted mobility. Therefore, reducing the human mobility is an essential strategy at the early stage of the pandemic.

However, individuals in different areas may practice the stay-at-home orders or travel restrictions in different ways. Despite the importance of the quarantine enforcement and contact tracing suggested by the United States Centers for Disease Control and Prevention (CDC), their implementations vary widely by the communities across the nation. Therefore, measuring the actual reductions in social contacts and travels plays a critical role in terms of understanding how governments evaluate the effectiveness of disease control policies. It is important to follow the CDC public health guidance for COVID-19 because the coronavirus is primarily transmitted through person-to-person contact when an infected individual coughs, sneezes, or talks, hence, mobility data has been used as a powerful tool to identify potentially exposed people. A significant correlation between the growth rate of COVID-19 and the mobility pattern has been examined in the United States (Badr et al., 2020). The effectiveness of household quarantine and contact tracing has been evaluated by exploring mobility data on the second wave of COVID-19 (Aleta et al., 2020). Awareness of following the CDC guidance, on the other hand, can improve the effectiveness of COVID-19 preventions, which has been confirmed with a research poll indicating a large percentage of population is taking precautions to maintain distance from each other (Saad, 2020). However, while research scientists can provide insights of the relationship between mobility and the coronavirus, it is still challenging to measure the efficacy of social distancing practices in terms of the ongoing pandemic with the difficulties of obtaining useful information.

Understanding human mobility has become a popular topic in urban science and smart city objects, as investigated by a wide range of research (Zhao et al., 2016; Wang et al., 2021). The data collection of the human mobility is the foundation of the whole development cycle of the decision-making process, which resorts to a set of social sensing and monitoring technologies that relevant data are collected from biological features, devices, and social media channels (Zhou et al., 2018; Wang & Taylor, 2016; Furini & Montangero, 2018). Visualizing and analyzing human mobility become a modern data-driven solution, as witnessed by many research topics via social sensing, ranging from visual mining of individual-level mobility patterns (Gonzalez et al., 2008), to urban planning (Yuan et al., 2012), including traffic congestions (Xu et al., 2016), sustainability problems (Prandi et al., 2017), and disease preventions (Charu et al., 2017). Such studies and applications provide rich experiences for exploring the human mobility against the COVID-19 pandemic. An early research is already started to examine how to track the COVID-19 infection pattern from mobility data with discussions on the effect of disease control policies (Kraemer et al., 2020).