Eugenia Nissi

The topic of this paper regards recent extensions of spatial scan statistics, widely used in public health research to test disease clusters and to identify their approximate locations. Despite its success, there is an important limitation associated with the traditional scan statistics: it depends on the use of circle shaped windows. As results, the identified regions are often not well localized. This limitation has motivated research aimed at developing new approaches which have the capability to detect clusters of irregular shapes. Two new techniques have been studied and compared: the spatial scan statistics, based on the graph theory, and the flexible scan statistics which imposes an irregularly shaped window. A computational study has been carried out to evaluate the effectiveness of these new approaches. A better understanding of the relative strengths and weakness of these two methods is essential to appropriate choices of methodology. Keywords: Detection Cluster Methods; Health Surveillance; Monte Carlo Testing; Simulated Annealing Scan Statistic; Flexible Scan Statistic.