ABSTRACT
The main purposes of this chapter are to summarize the current position of surveillance methods in public health and to focus on the inferential part of the surveillance problem including statistical techniques and stochastic modeling for evaluating surveillance systems engineering. At the same time, we provide an organized set of issues as a potential research roadmap, encouraging the ongoing research to implement the mechanism of change-point analysis for the detection of epidemics. Further, we discuss some of the statistical issues involved in the evaluation and optimal selection among change-point analysis-based approaches for early and accurate outbreak detection. The comparative study provides evidence that statistical change-point analysis-based methods have several appealing properties compared to the current practice. On the one hand, a Phase I distribution-free change-point analysis method is able to guarantee a prescribed false alarm probability without any knowledge about the (in-control) underlying distribution, whereas a periodic auto-regressive moving average model in conjunction with change-point detection seems to provide a useful tool to identify and model outbreaks that may occur in incidence data; thus, it can be proved beneficial to the society due to the consequences associated with the early detection and prevention of any type of extreme, possibly harmful, events.
