Irregular Spatial Cluster Detection Based on H1N1 Flu Simulation in Beijing

Spatial cluster detection of infected areas is widely used for disease surveillance, prevention and containment. However, the commonly used cluster methods cannot resolve the conflicts between the accuracy and efficiency of detection. We present an improved method for flexibly shaped spatial scanning, which can identify Irregular spatial clusters much more accurately and efficiently. First, we convert geographic information to a graph structure. Next, we approximately locate the disease regions. And then, based on the approximately located regions, we detect arbitrarily shaped and connected clusters in the graph based on likelihood ratio. Finally, we check the significance of the identified regions by Monte Carlo method. The algorithm is tested by an agent based simulation of H1N1 influenza data in Beijing. The results show that compared with the previous spatial scan statistic algorithms, our algorithm performs better in terms of shorter time and higher accuracy.