ABSTRACT
The field of spatio-temporal epidemiology has expanded rapidly in the past 10 years due to the development of statistical techniques that can accommodate variation over both space and time and the increasing availability of high-resolution data measuring a wide variety of environmental processes. Conventional methods for performing Bayesian analysis may be infeasible due to their high computational demands, paving the way for approximate methods for Bayesian inference such as INLA. The spatial field of residuals after subtracting an estimate of the mean proved highly non-stationary, so spatial deformation was used. Spatial deformation had a critical role to play. Dimension expansion has similarities with spatial deformation, but it differs in that the locations in the geographic space are retained, with added flexibility obtained through the extra dimensions.
