Analyses of infectious disease data with attention to heterogeneity
This work comprises three projects that extend previous models to include features of practical significance for the statistical analysis of infectious disease data. In the first, we find from a simulation study how the degree of heterogeneity in the number contacts that individuals have affects the relationship between estimates of a pathogen's effective population size based on coalescent theory and the true prevalence and incidence of that pathogen. In the second, we find that aggregating data from many small outbreaks allows the parameters of stochastic epidemic models to be consistently estimated with a generalized linear model. Application of this method to a set of 77 small norovirus outbreaks reveals interesting differences in the transmission parameters between hospital and nursing-home outbreaks. In the third project, we gain insight into HIV contact networks in the United States by fitting data from a number of surveys to a simple stochastic model of a dynamic network.