A1026
Title: Optimizing spatial cluster detection in small population: A comparative study of smoothing techniques
Authors: Chi Hsiung Lien - National Chengchi University (Taiwan) [presenting]
Yin Yee Leong - Feng Chia University (Taiwan)
Abstract: SatScan is a method for detecting spatial clusters that has been widely applied in many issues. When applied to small area populations, a small sample size may result in unstable clustering detection using SatScan. The issue of insufficient sample size is addressed in small area populations by employing smoothing methods such as partial SMR and Whittaker-Henderson before conducting clustering detection with SatScan, aiming to enhance the stability of detection. Simulation results demonstrate that the use of smoothing methods indeed improves the stability of detection. The extent of improvement varies with different smoothing methods under different clustering scenarios, with Whittaker-Henderson consistently showing stable improvement across all scenarios.
