A0421
Title: Reliability evaluation of regions within hotspot clusters using hierarchical structure of spatial data
Authors: Yusuke Takemura - Kyoto Women's University (Japan) [presenting]
Fumio Ishioka - Okayama University (Japan)
Koji Kurihara - Kyoto Women's University (Japan)
Abstract: In data such as the observed number of deaths due to infectious diseases in each region, there can be areas where the mortality risk is significantly higher than in the surrounding areas. These areas are referred to as hotspot clusters. Several methods utilizing spatial scan statistics have been proposed to detect clusters, and these methods have been widely used in fields such as epidemiology. However, in cluster detection using spatial scan statistics, there are regions that are erroneously detected as clusters due to slight fluctuations in observed values. Therefore, it is important to evaluate whether the regions detected as clusters should truly be included in clusters. To address this issue, the focus is on echelon analysis, a method for representing the hierarchical structure of spatial data. This method topologically classifies each region based on univariate values such as mortality rates, positioning regions with higher values in the higher hierarchy. A method is introduced for evaluating reliability by clarifying where regions included in detected clusters are located within the hierarchical structure obtained through echelon analysis.
