|
Activity Number:
|
16
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Sunday, August 8, 2004 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #300545 |
|
Title:
|
A Flexible Scan Statistic for Detecting Arbitrarily Shaped Clusters
|
|
Author(s):
|
Toshiro Tango*+ and Kunihiko Takahashi
|
|
Companies:
|
National Institute of Public Health, Japan and National Institute of Public Health, Japan
|
|
Address:
|
2-3-6 Minami, Wako, International, 351-0197, Japan
|
|
Keywords:
|
hot-spot clusters ; likelihood ratio test ; Poisson distribution ; spatial epidemiology ; power
|
|
Abstract:
|
Kulldorff's spatial scan statistic has been applied in a wide variety of epidemiological studies for cluster detection. His scan statistic, however, uses a circular window to define the potential cluster areas and thus has a difficulty of correctly identifying actual noncircular clusters. We propose a flexible scan statistic that can identify clusters of any shape. The performance of the proposed procedure is compared with that of Kulldorff's circle-based scan statistic via Monte Carlo simulation. We considered several circular and noncircular hot-spot cluster models and examined the newly introduced bivariate power distribution classified by the number of regions detected as the most likely cluster and the number of hot-spot regions (assumed in the simulation) included in the most likely cluster. The proposed flexible scan statistic is shown to have higher usual powers and also to identify the noncircular hot-spot clusters more accurately than Kulldorff's scan statistic. The proposed procedure is illustrated with some disease maps simulated in the Tokyo Metropolitan area.
|
