K-scan for anomaly detection in disease surveillance
Time and place
1 PM on Tuesday, March 17th, 2009; NAC 4/113
Prof. Ji Meng Loh (Columbia University)
Abstract
Anomaly detection in disease surveillance has in recent years been the focus of active research. Often, this involves searching for clusters of unusually high incidence rates among current cases of disease incidence, against a background or baseline incidence rate which may be spatially varying due to underlying variation in population density, or other population and environmental characteristics. We introduce a method, which we call K-scan, to identify such anomalies or hotspots. The procedure uses components of the inhomogeneous K function, often used to describe the clustering properties of spatial point data. Specifically, we assign to each case i a value K_i which, when summed together over all case locations, yields the overall inhomogeneous K function for the point pattern of disease locations. Points with high values of K_i are identified as potential members of clusters. We will present some results from a simulation study, as well as results from applying K-scan to dead bird sighting data from Contra Costa county of California.