Skip to main content

METHODS article

Front. Public Health
Sec. Infectious Diseases: Epidemiology and Prevention
Volume 12 - 2024 | doi: 10.3389/fpubh.2024.1432645
This article is part of the Research Topic Cluster-based Intelligent Recommendation System for Hybrid Healthcare Units View all 20 articles

Comparing Circular and Flexibly-shaped Scan Statistics for Disease Clustering Detection

Provisionally accepted
Lina Wang Lina Wang 1Xiang Li Xiang Li 2*Zhengbin Zhang Zhengbin Zhang 3Haoxun Yuan Haoxun Yuan 1Pengfei Lu Pengfei Lu 2*Yaru Li Yaru Li 1*
  • 1 School of Computer Science and Technology, Zhengzhou University of Light Industry, Zhengzhou, China
  • 2 Institute of Surveying and Mapping, PLA Information Engineering University, Zhengzhou, China
  • 3 Tuberculosis Prevention and Control Office, Other, Wuhan, China

The final, formatted version of the article will be published soon.

    The accuracy of spatial clustering detection is crucial for public health policy development and identifying etiological clues. Circular and flexibly-shaped scan statistics are widely used for disease cluster detection, but differences in results arise mainly due to parameter sensitivity and variations in the scanning window shapes. This study aims to analyze the impact of parameter settings on the results of these methods and compare their performance in disease clustering detection. Using tuberculosis data from Wuhan, China (2015-2019), the study identified the optimal parameter settings-MSWS and K-value-for each method to ensure accurate clustering. A comprehensive comparison was made using two quantitative indicators, the LLR value and cluster size, as well as clustering visualizations. The results show that the optimal MSWS parameter for SaTScan is determined through a Gini coefficient-based stepwise-threshold-reduction approach, while a K-value of 30 is ideal for FleXScan. SaTScan tends to produce more regular clusters, while FleXScan often generates more irregular clusters. FleXScan detects fewer clusters but with higher LLR values and larger average cluster sizes, although the maximum cluster size is smaller. These findings provide valuable insights for optimizing disease clustering detection methods and enhancing public health interventions.

    Keywords: Spatial scan statistics, Disease cluster detection, SaTScan, FlexScan, Gini Coefficient, log-likelihood ratio (LLR), cluster size

    Received: 14 May 2024; Accepted: 26 Dec 2024.

    Copyright: © 2024 Wang, Li, Zhang, Yuan, Lu and Li. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

    * Correspondence:
    Xiang Li, Institute of Surveying and Mapping, PLA Information Engineering University, Zhengzhou, China
    Pengfei Lu, Institute of Surveying and Mapping, PLA Information Engineering University, Zhengzhou, China
    Yaru Li, School of Computer Science and Technology, Zhengzhou University of Light Industry, Zhengzhou, China

    Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.