Spatiotemporally targeted lockdown as an alternative non-medical pandemic management measure – simulating a scattered accumulation as a result of localized closures
Description
Real-time tracking of the spatial diffusion of contact-based diseases, especially COVID-19, is a crucial point of recent academic studies and policymaking. Mobility networks provide a useful approach to grasp the complex and interrelated nature of spatiotemporal movement and the geographical spread of infectious diseases. In this simulation, we aimed at disclosing the spatial component of infection carried over by interpersonal encounters. The mobile call detail record (CDR) dataset we applied here is especially suited to estimate the number of interpersonal encounters enabled by co-locating the same space with an infected person within a definite timeframe. Based on the dataset, we elaborated a real-time lockdown model which can substantiate the smart urban pandemic management that fits into the smart city concept. A simulation was elaborated to prove that our model is suitable to scatter crowdedness around a central place and decrease personal exposure to infection. As our raster-level simulation showed that a wider area around the target site could be denoted with higher contamination risk, neighbourhoods are advised to be the spatial scale of smart intervention, which contributes to keeping infection risks at a manageable level. However, static visualisation can hardly represent the spatiotemporal dynamics that is taken place in the compelling interaction of smart closure at a central site and enhancing crowdedness at the boundaries. To overcome this, we elaborated a raster-rendered dynamic model throughout the day. The model addresses issues of real-time mobility sensing, spatiotemporally targeted actions, responsiveness to community problems and local resilience.
Files
Steps to reproduce
Our demonstration is intended to test real-time co-location analytics as a substitutional tool for pandemic management. The added value of our experiment is that we test a solution designed to use instead of complete lockdown and not an addition to that. This experiment demonstrates that temporally targeted and spatially focused lockdown disperses congestion all around the site and, by doing this, contributes to keeping infection risks at a manageable level. Four different stages of mobility restrictions were examined, according to the severity of closure instruction. Stage 1 refers to the baseline situation when no restrictions occur, and human mobility is free to flow. Stage 2 and Stage 3 describes careful lockdowns when mobility restrictions refer to the governmental recommendations with varying severity to persuade citizens to engage in self-limiting mobility behaviour. Stage 4 proposes full closures when individuals can only enter into the closed location when somebody left to secure public safety. In Stage 1, human mobility is free of flow, meaning that all individuals (100%) are allowed to enter the site. In Stage 2 and Stage 3, because of the governmental recommendation and citizens' self-limiting mobility choices, a share of 70% and 30% of baseline (Stage 1) human mobility flow enter the target location. Stage 4 prohibits entry into the area when the alarm threshold is reached, which implies that 100% of people are to be reversed from the bounded area, regardless of whether they arrive by public transport or are engaged in road or pedestrian traffic. According to our algorithm, if the number of co-located users reaches 300 in a specific raster within 15 minutes, the location gets labelled as closed (in Stage 2, 3 and 4). That implies that all users who would otherwise be logged in there would be diverted. A user who is not allowed to enter the location is assigned to one of the rasters at the closed area's boundary. She/he remains there until another log has been reported from another location visited by her/him later on the day. By that time, the individual leaves the site and appears in the other telecommunication tower's surroundings. Users who get inactive during the day may lower the descriptive power of the model. Those who got inactive more than one hour were dropped out from the sample to exclude uncertainties derived from scarce telecommunication events. The computation sequence of infection consists of a logical classification of 'susceptible', 'infected' and 'diseased' individuals. The necessary condition to become infected is being co-located with an infected person for at least 10 minutes. An individual can become susceptible and then infected and diseased later on. Computed progress of the disease is based on manipulating rank numbers that are attributed to susceptible and infected. Susceptible can become infected, and infected can become diseased based on sorting rank numbers.