B1315
Title: Improving the computational efficiency of spatial disease transmission models via clustering
Authors: Rob Deardon - University of Calgary (Canada) [presenting]
Abstract: Individual-level models (ILMs) of disease transmission incorporate individual-level covariate information, such as spatial location, to model infectious disease transmission. However, fitting these models with traditional Bayesian methods becomes cumbersome as model complexity or population size increases. Several methods have been proposed for reducing this computational burden by working on aggregated data obtained via spatial clustering algorithms. The aim is to review one such approach, the so-called cluster-aggregation-disaggregation algorithm, and discuss an alternative, novel approach based on spatially-composite analyses. In the former, aggregated data are treated as the new "individual-level" unit. In the latter, larger clusters are defined, and transmission between and within clusters is modelled using mechanisms that facilitate faster likelihood computation. The effectiveness of these methods is illustrated using simulated data and data from the UK 2001 foot-and-mouth disease epidemic.
