Spatial data aggregation for spatio-temporal individual-level models of infectious disease transmission. (May 2016)

• Record Type: Journal Article
• Authors: Deeth, Lorna E.; Deardon, Rob
• Is Part Of: Spatial and spatio-temporal epidemiology. Volume 17(2016)
• Page Start: 95
• Page End: 104
• Publication Date: 2016-05
• Subjects: Spatial clustering -- Disease modeling -- Individual-level models; Epidemiology -- Statistical methods -- Periodicals; Epidemiology -- Periodicals; 614.4072
• DOI: 10.1016/j.sste.2016.04.013
• Languages: English
• ISSNs: 1877-5845

Abstract

Highlights: Investigated effect of spatial data aggregation on identifying covariate effect. Simulations investigate reducing the computational expense via spatial data aggregation. Infectious disease models fitted to aggregated data set. Statistics used to determine if covariate effect can still be identified. Results suggest covariate effect can still be estimated under aggregated data. Abstract: A class of complex statistical models, known as individual-level models, have been effectively used to model the spread of infectious diseases. These models are often fitted within a Bayesian Markov chain Monte Carlo framework, which can have a sig nificant computational expense due to the complex nature of the likelihood function associated with this class of models. Increases in population size or duration of the modeled epidemic can contribute to this computational burden. Here, we explore the effect of reducing this computational expense by aggregating the data into spatial clusters, and therefore reducing the overall population size. Individual-level models, reparameterized to account for this aggregation effect, may then be fitted to the spatially aggregated data. The ability of two reparameterized individual-level models, when fitted to this reduced data set, to identify a covariate effect is investigated through a simulation study.