Research output: Contribution to Journal/Magazine › Journal article › peer-review
Research output: Contribution to Journal/Magazine › Journal article › peer-review
}
TY - JOUR
T1 - Efficient likelihood-free Bayesian Computation for household epidemics
AU - Neal, Peter
PY - 2012/11
Y1 - 2012/11
N2 - Considerable progress has been made in applying Markov chain Monte Carlo (MCMC) methods to the analysis of epidemic data. However, this likelihood based method can be inefficient due to the limited data available concerning an epidemic outbreak. This paper considers an alternative approach to studying epidemic data using Approximate Bayesian Computation (ABC) methodology. ABC is a simulation-based technique for obtaining an approximate sample from the posterior distribution of the parameters of the model and in an epidemic context is very easy to implement. A new approach to ABC is introduced which generates a set of values from the (approximate) posterior distribution of the parameters during each simulation rather than a single value. This is based upon coupling simulations with different sets of parameters and we call the resulting algorithm coupled ABC. The new methodology is used to analyse final size data for epidemics amongst communities partitioned into households. It is shown that for the epidemic data sets coupled ABC is more efficient than ABC and MCMC-ABC.
AB - Considerable progress has been made in applying Markov chain Monte Carlo (MCMC) methods to the analysis of epidemic data. However, this likelihood based method can be inefficient due to the limited data available concerning an epidemic outbreak. This paper considers an alternative approach to studying epidemic data using Approximate Bayesian Computation (ABC) methodology. ABC is a simulation-based technique for obtaining an approximate sample from the posterior distribution of the parameters of the model and in an epidemic context is very easy to implement. A new approach to ABC is introduced which generates a set of values from the (approximate) posterior distribution of the parameters during each simulation rather than a single value. This is based upon coupling simulations with different sets of parameters and we call the resulting algorithm coupled ABC. The new methodology is used to analyse final size data for epidemics amongst communities partitioned into households. It is shown that for the epidemic data sets coupled ABC is more efficient than ABC and MCMC-ABC.
KW - Approximate Bayesian Computation
KW - Household epidemic data
KW - Stochastic epidemic models
U2 - 10.1007/s11222-010-9216-x
DO - 10.1007/s11222-010-9216-x
M3 - Journal article
VL - 22
SP - 1239
EP - 1256
JO - Statistics and Computing
JF - Statistics and Computing
SN - 0960-3174
IS - 6
ER -