Bayes Factors for Partially Observed Stochastic Epidemic Models

We consider the problem of model choice for stochastic epidemic models given partial observation of a disease outbreak through time. Our main focus is on the use of Bayes factors. Although Bayes factors have appeared in the epidemic modelling literature before, they can be hard to compute and little attention has been given to fundamental questions concerning their utility. In this paper we derive analytic expressions for Bayes factors given complete observation through time, which suggest practical guidelines for model choice problems. We extend the power posterior method for computing Bayes factors so as to account for missing data and apply this approach to partially observed epidemics. For comparison, we also explore the use of a deviance information criterion for missing data scenarios. The methods are illustrated via examples involving both simulated and real data.

[1]  H. Andersson,et al.  Stochastic Epidemic Models and Their Statistical Analysis , 2000 .

[2]  Martin Eichner,et al.  Transmission potential of smallpox: estimates based on detailed data from an outbreak. , 2003, American journal of epidemiology.

[3]  N. C. Severo.  Generalizations of some stochastic epidemic models , 1969 .

[4]  P. Yip,et al.  ANALYSIS OF VARIATIONS IN AN INFECTION RATE , 1989 .

[5]  Philip D O'Neill,et al.  Bayesian model choice for epidemic models with two levels of mixing. , 2014, Biostatistics.

[6]  P. O’Neill,et al.  Inference for Epidemics with Three Levels of Mixing: Methodology and Application to a Measles Outbreak , 2011 .

[7]  Nial Friel,et al.  Improving power posterior estimation of statistical evidence , 2012, Stat. Comput..

[8]  K. Atkinson.  Elementary numerical analysis , 1985 .

[9]  G. Gibson,et al.  New model diagnostics for spatio-temporal systems in epidemiology and ecology , 2014, Journal of The Royal Society Interface.

[10]  R. Deardon,et al.  Model Choice Using the Deviance Information Criterion for Latent Conditional Individual-Level Models of Infectious Disease Spread , 2015 .

[11]  A. Pettitt,et al.  Marginal likelihood estimation via power posteriors , 2008 .

[12]  P. O’Neill,et al.  Bayesian model choice and infection route modelling in an outbreak of Norovirus , 2005, Statistics in medicine.

[13]  H. Philippe,et al.  Computing Bayes factors using thermodynamic integration. , 2006, Systematic biology.

[14]  P D O'Neill,et al.  Modelling and inference for epidemic models featuring non-linear infection pressure. , 2012, Mathematical biosciences.

[15]  C. Robert,et al.  Deviance information criteria for missing data models , 2006 .

[16]  G. Roberts,et al.  Bayesian inference for partially observed stochastic epidemics , 1999 .

[17]  Lin Zhang.  Time-varying Individual-level Infectious Disease Models , 2013 .

[18]  Theodore Kypraios,et al.  Bayesian non-parametric inference for stochastic epidemic models using Gaussian Processes , 2016, Biostatistics.

[19]  Gavin J. Gibson,et al.  Estimating parameters in stochastic compartmental models using Markov chain methods , 1998 .

[20]  Xiao-Li Meng,et al.  Simulating Normalizing Constants: From Importance Sampling to Bridge Sampling to Path Sampling , 1998 .

[21]  G. Roberts,et al.  Statistical inference and model selection for the 1861 Hagelloch measles epidemic. , 2004, Biostatistics.

[22]  Theodoros Kypraios,et al.  Efficient Bayesian inference for partially observed stochastic epidemics and a new class of semi-parametric time series models , 2007 .

[23]  C. Worby.  Statistical inference and modelling for nosocomial infections and the incorporation of whole genome sequence data , 2013 .

[24]  Theodore Kypraios,et al.  Modelling and Bayesian analysis of the Abakaliki smallpox data. , 2016, Epidemics.

[25]  Muteb Alharthi.  Bayesian model assessment for stochastic epidemic models , 2016 .

[26]  P. Neal,et al.  Model comparison with missing data using MCMC and importance sampling , 2015, 1512.04743.

[27]  Xiaoguang Xu,et al.  Bayesian nonparametric inference for stochastic epidemic models , 2015 .