Geo-visualization and Clustering to Support Epidemiology Surveillance Exploration

Abstract

WebEpi is an epidemiological WebGIS service developed for the Population Health Epidemiology Unit of the Tasmania Department of Health and Human Services (DHHS). Epidemiological geographical studies help analyze public health surveillance and medical situations. It is still a challenge to conduct large-scale geographical information exploration of epidemiology surveillance based on patterns and relationships. Generally, there are two crucial stages for GIS mapping of epidemiological data: one precisely clusters areas according to their health rate, the other efficiently presents the clustering result on GIS map which aims to help health researchers plan health resources for disease prevention and control. There are two major cluster algorithms for health data exploration, namely Self Organizing Maps (SOM) and K-means. In this paper, the clustering based on SOM and K-means are presented and their clustering results are compared by their clustering process and mapping results. It is concluded from experimental results that K-means produces a more promising mapping result for visualizing the highest mortality rate municipalities.