Clustering with spatial constraints: The case of diarrhea in Bandung city, Indonesia

Abstract

Clustering with spatial constraints was introduced in disease surveillance to identify high/low risk areas. These improve the geographical pattern of disease clustering. There are two types of spatial constraints, geographical distance and contiguity based. A Ward-like hierarchical clustering algorithm, including the spatial constraints, was employed on diarrheal data in Bandung city. Diarrhea is an infectious disease that causes death. The geographical distance was the best geographical dissimilarity for the diarrheal disease data. A five-cluster solution was determined to be optimal. Cluster two, which consists of three districts (Cibiru, Cinambo, Mandalajati), was considered as a high-risk cluster. It scored high on standardized incidence ratio which is caused by low on healthy house index and water quality index.