Bayesian inference for multi-strain epidemics with application to Escherichia coli O157:H7 in feedlot cattle
Panayiota Touloupou (University of Warwick)
Friday 28th February, 2020 15:00-16:00 Maths 311B
For most pathogens there exist testing procedures that can be used to distinguish among dierent strains with which individuals are infected. Due to the growing availability of such data, multi- strain models have increased in popularity over the past years. Statistical inference for such models is not trivial and requires specialised methodology. A key facet of the problem is that the times of acquiring and clearing infection are not directly observed and these are required to calculate the likelihood function on which inferences are based. The usual solution is to treat these times as missing data within a Bayesian framework and impute them using Markov chain Monte Carlo techniques. When multiple strains of a pathogen are coexisting then the number of infectious states an individual can exhibit is greatly increased and existing inference approaches become prohibitive.
Motivated by this demand, we construct a model that incorporates this additional information regarding the strains in which the bacterium appears and propose a statistical algorithm for infer- ence. The model improves upon existing methods in the sense that it allows for both imperfect diagnostic test sensitivities and strain misclassication. Extensive simulation studies are conducted in order to assess the performance of our method, while the utility of the developed methodology is demonstrated on data obtained from a longitudinal study of Escherichia coli O157:H7 in feedlot cattle in which 8 competing strains were identied using genetic typing methods. We show that surprisingly little genetic data is needed to produce a probabilistic reconstruction of the epidemic trajectories, despite some possibility of misclassication in the genetic typing. We believe that this complex model, capturing the interactions between strains, would not have been able to be tted using existing methodologies.