Fine-scale variation in the effect of national border on COVID-19 spread: A case study of the Saxon-Czech border region.

The global extent and temporally asynchronous pattern of COVID-19 spread have repeatedly highlighted the role of international borders in the fight against the pandemic. Additionally, the deluge of high resolution, spatially referenced epidemiological data generated by the pandemic provides new opportunities to study disease transmission at heretofore inaccessible scales. Existing studies of cross-border infection fluxes, for both COVID-19 and other diseases, have largely focused on characterizing overall border effects. Here, we couple fine-scale incidence data with localized regression models to quantify spatial variation in the inhibitory effect of an international border. We take as a case study the border region between the German state of Saxony and the neighboring regions in northwestern Czechia, where municipality-level COVID-19 incidence data are available on both sides of the border. Consistent with past studies, we find an overall inhibitory effect of the border, but with a clear asymmetry, where the inhibitory effect is stronger from Saxony to Czechia than vice versa. Furthermore, we identify marked spatial variation along the border in the degree to which disease spread was inhibited. In particular, the area around Löbau in Saxony appears to have been a hotspot for cross-border disease transmission. The ability to identify infection flux hotspots along international borders may help to tailor monitoring programs and response measures to more effectively limit disease spread.

PIETS: Parallelised Irregularity Encoders for Forecasting with Heterogeneous Time-Series

Heterogeneity and irregularity of multi-source data sets present a significant challenge to time-series analysis. In the literature, the fusion of multi-source time-series has been achieved either by using ensemble learning models which ignore temporal patterns and correlation within features or by defining a fixed-size window to select specific parts of the data sets. On the other hand, many studies have shown major improvement to handle the irregularity of time-series, yet none of these studies has been applied to multi-source data. In this work, we design a novel architecture, PIETS, to model heterogeneous time-series. PIETS has the following characteristics: (1) irregularity encoders for multi-source samples that can leverage all available information and accelerate the convergence of the model;(2) parallelised neural networks to enable flexibility and avoid information over-whelming;and (3) attention mechanism that highlights different information and gives high importance to the most related data. Through extensive experiments on real-world data sets related to COVID-19, we show that the proposed architecture is able to effectively model heterogeneous temporal data and outperforms other state-of-the-art approaches in the prediction task.

Synthetic People Flow: Privacy-Preserving Mobility Modeling from Large-Scale Location Data in Urban Areas

Recently, there has been an increasing demand for traffic simulation and congestion prediction for urban planning, especially for infection simulation due to the Covid-19 epidemic. On the other hand, the widespread use of wearable devices has made it possible to collect a large amount of user location history with high accuracy, and it is expected that this data will be used for simulation. However, it is difficult to collect location histories for the entire population of a city, and detailed data that can reproduce trajectories is expensive. In addition, such personal location histories contain private information such as addresses and workplaces, which restricts the use of raw data. This paper proposes Agent2Vec, a mobility modeling model based on unsupervised learning. Using this method, we generate synthetic human flow data without personal information. © 2022, ICST Institute for Computer Sciences, Social Informatics and Telecommunications Engineering.