This talk will give an overview of some recent advances on Bayesian methods for jointly modelling epidemic data with genomic data on pathogens within an integrated analysis. We will illustrate how latent processes can be used to tackle problems of model assessment and comparison in a way that avoids the complexities and sensitivities of a fully Bayesian approach. In particular we show how latent residual processes can be designed and tested, using classical ideas embedded in the Bayesian framework, in order to detect particular forms of model mis-specification. The talk will demonstrate how the methods can be used to assess assumptions regarding spatial infection kernels in epidemic models and assumptions regarding within-host diversity in the genetic component of a joint model, as applied to Foot and Mouth Disease in the UK.

This research has been conducted in collaboration with Max Lau, David Thong, Glenn Marion, George Streftaris, Colin Worby and Bryan Grenfell with funding from the Engineering and Physical Sciences Research Council,UK and Biomathematics and Statistics Scotland.