Deriving pandemic disease mitigation strategies by mining social contact networks


In this chapter we propose a robust approach to deriving disease mitigation strategies from contact networks that are generated from publicly available census data. The goal is to provide public policy makers additional information concerning the type of people they should aim to target vaccination, quarantine, or isolation measures towards. We focus on pandemic disease mitigation, but the approach can be applied to other domains, such as bioterrorism. The approach begins by constructing a representative contact network for the geographic area (we use the Greater Toronto Area of ≈ 5.5 million individuals) from census information. Then, network centrality measures are employed to ascertain the importance of each individual to the proper topological functioning of the network. The top-ranked individuals' characteristics, as defined by census information, are then used as input to decision tree classifiers. The resulting output is a set of high-level rules that identify potential types of individuals to target in order to mitigate disease spread. Experimental evidence for the efficacy of the approach is also provided.

Springer Proceedings in Mathematics and Statistics
Dionne M. Aleman, PhD, PEng
Dionne M. Aleman, PhD, PEng
Professor of Industrial Engineering