BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Bayesian experimental design for stochastic dynamical models - Gib
 son\, GJ (Heriot-Watt University)
DTSTART:20110721T103000Z
DTEND:20110721T113000Z
UID:TALK32118@talks.cam.ac.uk
CONTACT:Mustapha Amrani
DESCRIPTION:Advances in Bayesian computational methods have meant that it 
 is now possible to fit a broad range of stochastic\, non-linear dynamical 
 models (including spatio-temporal formulations) within a rigorous statisti
 cal framework.  In epidemiology these methods have proved particularly val
 uable for producing insights into transmission dynamics on historical epid
 emics and for assessing potential control strategies. On the other hand\, 
 there has been less attention paid to the question how future data should 
 be collected most efficiently for the purpose of analysis with these model
 s. This talk will describe how the Bayesian approach to experimental desig
 n can be applied with standard epidemic models in order to identify the mo
 st efficient manner for collecting data to provide information on key rate
  parameters. Central to the approach is the representation of the design a
 s a 'parameter' in an extended parameter space with the optimal design app
 earing as the marginal mode for an appropriately specified joint distribut
 ion. We will also describe how approximations\, derived using moment-closu
 re techniques\, can be applied in order to make tractable the computationa
 l of likelihood functions which\, given the partial nature of the data\, w
 ould be prohibitively complex using methods such as data augmentation. The
  talk will illustrate the ideas in the context of designing microcosm expe
 riments to study the spread of fungal pathogens in agricultural crops\, wh
 ere the design problem relates to the particular choice of sampling times 
 used. We will examine the use of utility functions based entirely on infor
 mation measures that quantify the difference between prior and posterior p
 arameter distributions\, and also discuss how economic factors can be inco
 rporated in the construction of utilities for this class of problems.  The
  talk will demonstrate how\, if sampling times are appropriately selected\
 , it may be possible to reduce drastically the amount of sampling required
  in comparison to designs currently used\, without compromising the inform
 ation gained on key parameters. Some challenges and opportunities for futu
 re research on design with stochastic epidemic models will also be discuss
 ed.\n
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR