Detection of space–time clusters using a topological hierarchy for geospatial data on COVID-19 in Japan

• Published in: Japanese journal of statistics and data science Vol. 5; no. 1; pp. 279 - 301
• Main Authors: Takemura, Yusuke, Ishioka, Fumio, Kurihara, Koji
• Format: Journal Article
• Language: English
• Published: Singapore Springer Nature Singapore 01.07.2022
• Subjects: Chemistry and Earth Sciences; Computer Science; Economics; Finance; Health Sciences; Humanities; Insurance; Law; Management; Mathematics and Statistics; Medicine; Original Paper; Physics; Statistical Theory and Methods; Statistics; Statistics and Computing/Statistics Programs; Statistics for Business; Statistics for Engineering; Statistics for Life Sciences; Statistics for Social Sciences; COVID-19; Spatial scan statistic; Echelon analysis; Space–time cluster
• ISSN: 2520-8756; 2520-8764
• DOI: 10.1007/s42081-022-00159-x

In this paper, we detected space–time clusters using data on coronavirus disease 2019 (COVID-19) collected daily by each prefecture in Japan. COVID-19 has spread globally since the first confirmed case in China, in December 2019. Several people have to date been infected in Japan since the first confirmed case in January 2020. The outbreak of COVID-19 has had a significant impact on many people’s lives. Studies are being conducted to detect regions, called clusters, which pose a significantly higher risk of infection than their surrounding areas, based on a spatial scan statistics of COVID-19 infections. Among these studies, space–time cluster detection has to date been actively performed to gain knowledge regarding infection status. Based on the spatial scan statistic, the cylindrical scan method is a widely used space–time cluster detection method. This method enables concurrent detection of the location and time of a cluster occurrence. However, this method cannot capture spatial changes in a cluster over time. When applying the existing method to a cluster whose shape changes over time, the number of calculations required becomes extremely large, and the analysis may become difficult. In this study, we focused on the hierarchical structure of the data obtained by conducting an echelon analysis and applied the space–time cluster detection method based on this structure to enable the capture of changes in a cluster’s shape. Furthermore, we visualized the location and period of a cluster’s occurrence and considered the cause of the cluster.