Scan Statistics for Space-Time Cluster Detection
Abstract: Scan statistics are used by public health agencies to detect and localize disease outbreaks. This thesis provides an overview of scan statistics in the context of prospective disease surveillance and outbreak detection, presents a novel scan statistic to deal with the type of zero-abundant data that is often encountered in these settings, and—perhaps most importantly—implements this and other scan statistics in a freely available and open source R package. Additionally, Markov processes and time series methods are frequently used in many disease surveillance methods. The last part of this thesis presents some computationally efficient methods for density evaluation and simulation of irregularly sampled AR(1) processes, that may be useful when implementing surveillance methods based on these types of processes.
This dissertation MIGHT be available in PDF-format. Check this page to see if it is available for download.