Award details

Experimental design for stochastic dynamical models in the life sciences

Reference	BB/C007263/1
Principal Investigator / Supervisor	Professor Gavin Gibson
Co-Investigators / Co-Supervisors	Professor Christopher Gilligan
Institution	Heriot-Watt University
Department	S of Mathematical and Computer Sciences
Funding type	Research
Value (£)	184,739
Status	Completed
Type	Research Grant
Start date	19/09/2005
End date	18/09/2009
Duration	48 months

Abstract

The project will use Bayesian techniques for optimally designing experiments on host-pathogen interactions in populations of humans, animals and plants. It will build on previous achievements of the applicants who fitted stochastic dynamical models to experimental data using computational methods. While the problems associated with fitting these models to partial observations have been overcome to some extent using Markov chain Monte Carlo (MCMC) methods, the problem of designing experiments to maximise information on model parameters has received less attention. The proposed approach extends that of data augmentation used in model fitting, whereby latent aspects of the process are treated as additional unknown parameters whose joint posterior distribution was investigated by MCMC. In the proposed project, this framework is extended to include future realisation, y, of the process and the design, d, as additional unknown components. By setting the conditional density of (y,d) to be proportional to f(y)U(d,y) where f denotes the predictive density of y and U(d,y) is a utility function, we find that the marginal distribution of the design is proportion to its expected utility. The framework, though potentially powerful, requires some substantial problems to be tackled. These include the specification of appropriate utility functions that can be readily computed or estimated, and the development of new MCMC algorithms that can be applied to estimate parameters for the sparser designs that will emerge. The techniques will be applied to historical data on the spread of citrus tristeza virus and citrus canker and in ongoing experimental programmes on fungal pathogen spread in microcosm populations. The range of applications therefore includes both explicitly spatio-temporal systems and models and non-spatial models. The advances in the project will have the potential to impact on studies of a broad range of host-pathogen systems.

Summary

unavailable

Committee	Closed Committee - Engineering & Biological Systems (EBS)
Research Topics	Animal Health, Crop Science, Microbiology, Plant Science, Systems Biology, Technology and Methods Development
Research Priority	X – Research Priority information not available
Research Initiative	X - not in an Initiative
Funding Scheme	X – not Funded via a specific Funding Scheme

I accept the terms and conditions of use (opens in new window)

export PDF file

back to list new search