Simulation of the geographical spread of COVID-19 disease based on mobile cell data: animated dispersal of undiagnosed infected and modelling the mobility behaviour under displacement restrictions

Real-time tracking of the spatial diffusion of airborne diseases, and especially COVID-19 is in the focal point of both recent academic studies and policymaking. Airborne pathogens are handed over by interpersonal encounters. Therefore, agent-based modelling provides a useful approach to grasp the complex and interrelated nature of spatiotemporal movement and the geographical spread of infectious diseases. Although technology development rendered it to be feasible to track the spatial spread of infected individuals, the spatial scale of data retrieval can cause challenging bottlenecks for academic analysis. Samples on community-scale, for instance, by crowdsourced data as well as the global level of international aircraft movements are addressed. However, regional-scale spread of airborne diseases conveyed by human mobility rarely comes into focus. By directing our efforts to the level of countrywide diffusion, we aim to disclose the spatial component of airborne pathogens’ infection carried over by interpersonal encounters. The mobile cell dataset we applied here is especially suitable to estimate the number of interpersonal encounters, that is enabled by co-locating the same space with an infected person within a definite timeframe. Consequently, we considered mobile phone data driven co-location as ‘locational chance’ of airborne pathogen spreading.
The volume of spread, as we argue, is dependent on the interpersonal connections. According to the current results, the geographical spread of COVID-19 is dominantly carried over by latently infected individuals, who transmit the disease without showing any symptoms. We modelled the interpersonal encounters of a set of randomly chosen latent infected as an indicator of the further geographical spread of the disease. We applied two various sets of models running: one, that is based on real archive data, and the other, that simulates current mobility patterns ordered by relocation restrictions.

实时监测空气中传播疾病的时空扩散，尤其是针对COVID-19的研究和政策制定，已成为学术界和决策层的关注焦点。空气传播的病原体通过人际接触传播。因此，基于智能体的建模为深入理解时空移动的复杂性和相互关联性，以及传染病的地理传播提供了有效的途径。尽管技术的发展使得追踪感染个体的空间传播成为可能，但数据检索的空间尺度可能成为学术分析的瓶颈。例如，社区尺度的样本，包括通过众包数据以及国际航空运动的全球层面，均被考虑在内。然而，由人类流动性引发的地区尺度空气中疾病的传播往往未受到重视。通过将我们的努力集中在国家层面的扩散上，我们旨在揭示人际接触携带的空气中病原体感染的空间成分。我们应用的可移动单元数据集特别适用于估计人际接触的数量，即在特定时间段内与感染者同处一地的可能性。因此，我们认为由移动电话数据驱动的共定位是空气中病原体传播的‘地理位置机遇’。正如我们所论证的，传播量依赖于人际联系。根据当前结果，COVID-19的地理传播主要是由潜伏感染者携带，他们在不显示任何症状的情况下传播疾病。我们将随机选择的一组潜伏感染者的人际接触建模为疾病进一步地理传播的指标。我们应用了两套不同的模型运行：一套基于真实档案数据，另一套模拟了由搬迁限制排序的当前流动性模式。