MCMC II for Infectious Diseases
Meeting Times:
- Wednesday, July 24, 1:30 PM – 5:00 PM
- Thursday July 25, 8:30 AM – 5:00 PM
- Friday July 26, 8:30 AM – 5:00 PM
Classroom: Randall Rollins Building (RR 200)
Module Summary:
This module looks in detail at practical implementation issues for MCMC methods when applied to data from infectious disease outbreaks. The main focus will be towards inference for the SIR (susceptible-infected-removed) model. Topics include parameterization, methods for improving convergence, assessing MCMC output, and data augmentation methods. Programming will be carried out in R.
Prerequisites:
The course assumes all the material in Module: MCMC I for Infectious Diseases or the equivalent knowledge of Bayesian inference (e.g. understanding prior, likelihood, posterior) and MCMC (e.g. understanding how to implement Gibbs sampling and Metropolis-Hastings algorithms).
Students are expected to have a working knowledge of the R computing environment. Programming will be in R. Students new to R should complete an extensive tutorial before the module.
Module Content:
- Simulation
- Stochastic SIR model: detailed explanation of fitting to partial observation data
- Household models: fitting to temporal and final size data
- More complex modeling examples
- Brief introduction to ABC methods
Instructors
Phil O’Neill, PhD
Professor of Applied Probability, University of Nottingham, UK
Phil’s research interests include Bayesian inference methods for the analysis of infectious disease data, methods for integrating epidemiological and genomic data, and the probabilistic analysis of stochastic epidemic models.
Theo Kypraios, PhD
Professor of Statistics, University of Nottingham, UK
Theo's research is concerned with the development of novel computational statistical methodology for Bayesian inference and model selection for high-dimensional complex data. The area that he has mostly worked on and made contributions to, is infectious disease modeling. His particular focus is on the design of efficient Monte Carlo methods (e.g. Markov Chain Monte Carlo, Sequential Monte Carlo and Approximate Bayesian Computation), including recent work in developing novel Bayesian Nonparametrics methodology for epidemic models.
Required Software: R
Recommended Reading:
Recommended, but not required: Knowledge of the material from the modules Mathematical Models of Infectious Diseases or Stochastic Epidemic Models with Inference, or the equivalent, would be helpful, but not required.
