< Back to Article

Agricultural and geographic factors shaped the North American 2015 highly pathogenic avian influenza H5N2 outbreak

Fig 2

Comparison of hypothesized HPAI H5N2 epidemiological compartmental models.

(A) Each compartmental model represents a Susceptible-Infectious-Removed (SIR) model with varied population heterogeneity: 1) a single, closed, homogenous population, 2) a single, homogenous population with a continual external source of virus (U), 3) a closed population, stratified by poultry system (turkeys (T) and layer chickens (C)), and 4) the stratified population with a continual external source of virus. (B) Compartmental model fit for the midwestern highly pathogenic avian influenza (HPAI) H5N2 outbreak, 2015. Akaike’s information criteria for Markov chain Monte Carlo (AICM) calculated based on the posterior distribution of the structured tree likelihood was used to evaluate the relative model fit for the four assessed compartmental models under differing molecular clock assumptions. Under both molecular clocks, Model 3 provided the best model fit. (C) Estimated infectious period of layer chicken and turkey farms during the 2015 midwestern highly pathogenic avian influenza (HPAI) H5N2 outbreak. During model specification, an informative prior was provided for the Bayesian process. This prior probability distribution was based on the reported average time from HPAI confirmation to depopulation plus 5 days to allow for delay between infection and HPAI confirmation. Model 3 estimated the infectious period for layer chickens to be longer than expected given the prior information.

Fig 2