An Early Detection System for Disease Outbreaks Based on Density Entropy

Cluster analysis in epidemiology and surveillance systems has made possible the early detection of the disease outbreak. This paper presents a disease outbreak detection system that includes a combination of Alpha shape method and an improved novel clustering approach based on density entropy. In this approach, the individual patient address collected by the disease surveillance system is first converted into planar coordinates, and then the cluster is given by our algorithm, the border is computed by the Alpha shape method. Lastly, the result is compared with the circular scan statistic on log likelihood ratio and individual density. The experimental results show the performance of our approach is better than the circular scan statistic especially when there have irregularly shaped disease outbreak areas within or across the administrative regions.