A scan statistic for continuous data based on the normal probability model

Abstract

Temporal, spatial and space-time scan statistics are commonly used to detect and evaluate the statistical significance of temporal and/or geographical disease clusters, without any prior assumptions on the location, time period or size of those clusters. Scan statistics are mostly used for count data, such as disease incidence or mortality. Sometimes there is an interest in looking for clusters with respect to a continuous variable, such as lead levels in children or low birth weight. For such continuous data, we present a scan statistic where the likelihood is calculated using the the normal probability model. It may also be used for other distributions, while still maintaining the correct alpha level. In an application of the new method, we look for geographical clusters of low birth weight in New York City.

A scoping review of spatial cluster analysis techniques for point-event data

“…In particular, our review found that the spatial scan statistic was utilized for spatial cluster detection in 19 of the 29 reviewed papers. Much has been discussed about the flexibility of the spatial scan statistic (Kulldorff, 1999;Kulldorff et al, 2006Kulldorff et al, , 2009Costa and Kulldorff, 2009), however, there has been equal amount of research highlighting some weaknesses of the method (Tango and Takahashi, 2005;Neill, 2009;Cançado et al, 2010 ). Recent uses in spatial clustering methods, GAMs (Vieira et al, 2009) and KIFs (Wheeler, 2007), illustrate the advantages of generating a smooth risk surface coupled with an intuitive visual output.…”

Functional Data Analysis for Optimizing Strategies of Cash-Flow Management

“…The purpose of this study is to ascertain where and when the high contaminant cluster lies at the "difficult-to-return zone" in Fukushima prefecture from these monitoring results. The spatial scan statistic (Kulldorff et al, 2009) is a method of detecting clusters based on the likelihood ratio associated with the quantity inside and outside a circular scanning window. However, it is noted that a non-circular shaped cluster, such as the shape made by a river or a road cannot be detected.…”

New developments in the spatial scan statistic

“…elliptical [24], echelon-based [25] or free-form [26].Following Kulldorff's derivation of the Bernoulli and Poisson SSSs (for case-control point data and areal count data, respectively), many other versions have been proposed, e.g. for normally distributed data [27], survival data [28], and ordinal data [29]. Less attention has been paid to part 2 of the algorithm, from which much of the computational cost of the process arises.…”