Bayesian parameter inference in discrete state-space stochastic population models

Continuous-time discrete state-space Markov processes are useful dynamic models for a wide range of systems in biology. These include wildlife populations, chemical kinetics (in small volumes), animal behaviour, and the spread of disease in space and time. In some cases model parameters can be determined from targeted experiments, but in many cases appropriate data is not available, or model parameters can't be determined independently. It is then desirable to determine combined parameter values from direct/global observations of the system. In this talk I will outline a Bayesian approach to parameter inference from incomplete data as applied to such models and systems. I will attempt to provide a genuine introduction to the topic: (i) outlining the Bayesian approach and why it is a natural choice for such problems; (ii) describing the mechanics of applying the approach; and (iii) describing some examples and associated practical problems.