University of Cambridge > > Worms and Bugs > Exact simulation-based Bayesian inference for epidemic models

Exact simulation-based Bayesian inference for epidemic models

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact Prof. Julia Gog.

Inference in epidemic models poses many challenges, not least because of missing or unobserved data. A powerful method for tackling some of these issues is to use a Bayesian framework and data-augmented Markov chain Monte Carlo fitting algorithms. However, these techniques can become computationally intensive for large-scale systems. An alternative is to use pseudo-marginal algorithms (O’Neill et al., 2000; Beaumont, 2003; Andrieu and Roberts, 2009), which provide methods for estimating both exact and approximate posterior distributions for the parameters-of-interest based on importance sample estimates generated from model simulations. When the observation process is deterministic, then this requires that the model simulations match the observed data exactly, which can be problematic in highly stochastic systems without the availability of large amounts of computing power. We present some methods for reducing stochasticity and improving computational efficiency for simulations of epidemic models, by conditioning the simulations on the model and data. We illustrate these techniques on real data for a variety of model/data combinations.

This talk is part of the Worms and Bugs series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.


© 2006-2024, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity