An Examination of Three Spatial Event Cluster Detection Methods

ISPRS International Journal of Geo-Information

Hensley H. Mariathas²  Rhonda J. Rosychuk¹ 

[1] Department of Pediatrics, University of Alberta, Edmonton, Alberta, T6G 2J3, Canada; E-Mail

关键词: spatial event cluster;    cluster detection;    compound Poisson distribution;    approximate normal distribution;    multiple hypergeometric distribution;    surveillance;    substance use;   

DOI  :  10.3390/ijgi4010367

来源: mdpi

【 摘 要 】

In spatial disease surveillance, geographic areas with large numbers of disease cases are to be identified, so that targeted investigations can be pursued. Geographic areas with high disease rates are called disease clusters and statistical cluster detection tests are used to identify geographic areas with higher disease rates than expected by chance alone. In some situations, disease-related events rather than individuals are of interest for geographical surveillance, and methods to detect clusters of disease-related events are called event cluster detection methods. In this paper, we examine three distributional assumptions for the events in cluster detection: compound Poisson, approximate normal and multiple hypergeometric (exact). The methods differ on the choice of distributional assumption for the potentially multiple correlated events per individual. The methods are illustrated on emergency department (ED) presentations by children and youth (age < 18 years) because of substance use in the province of Alberta, Canada, during 1 April 2007, to 31 March 2008. Simulation studies are conducted to investigate Type I error and the power of the clustering methods.

【 授权许可】

CC BY   
© 2015 by the authors; licensee MDPI, Basel, Switzerland