A1259
Title: Application of a reliability evaluation method for space-time clusters using a hierarchical structure
Authors: Yusuke Takemura - Kyoto Women's University (Japan) [presenting]
Fumio Ishioka - Okayama University (Japan)
Koji Kurihara - Kyoto Women's University (Japan)
Abstract: In spatial epidemiology, spatiotemporal data containing information on the locations and times where infected individuals were observed are actively released for infectious diseases such as COVID-19. Space-time clusters with information on location and duration of occurrence have been detected for various spatiotemporal datasets. An echelon scan method that utilizes the spatial hierarchical structure of data for detecting space-time clusters was developed. The method can capture changes in cluster shape, such as expansion or division, which are difficult to detect using other methods. However, the method is sensitive to random fluctuations in observed values, making the assessment of cluster reliability a critical issue in practical applications. Since the true shape of clusters is unknown, it is difficult to determine whether the regions within a detected cluster genuinely represent elevated risk or are simply the result of random variation. An approach for evaluating the reliability of detected clusters while explicitly accounting for uncertainty in the observed data is introduced.
