ABSTRACT
This chapter is concerned with kernel methods for examining the spatial distribution and variation in disease risk for individual event data. Specifically, we assume that the geographical coordinates of disease cases (and perhaps controls) are observed. This may occur when we have very precise information on these locations (e.g., the home addresses of individuals), but the methods are also generally applicable to situations in which cases are identified within spatial units that are tiny in comparison to the study region as a whole. In such circumstances, it is typically convenient to model the data as points located at the centroids of these small areas.
