Spatio-Temporal Surveillance Using CUSUM of Order Statistics

The scan statistic is a popular choice for monitoring and detecting spatio‐temporal outbreaks. It can be designed to be optimal if the outbreak characteristics (shape and size) are known in advance. However, in all practical situations, neither the shape nor the size are known in advance. Therefore, there is a need for spatio‐temporal surveillance plans that perform well for a range of unknown outbreaks. This paper proposes a new approach for detecting spatio‐temporal outbreaks based on the cumulative sum of order statistics. The approach performed on average better than the simple scan statistic for both a range of outbreaks involving a single geographical region. More importantly, it performed significantly better than the simple scan plan for outbreaks involving simultaneous multiple (non‐overlapping) geographically dispersed regions.