Title: Bayesian approaches in Microbial Risk Assessment
Speaker: Helen Clough (Liverpool University)
Venue: Fylde SCR, Lancaster university.
Date: 2010-06-24(Thurs) 16:00
All welcome!
Ting-Li Su
Secretary,RSS Lancashire/Cumbria Local Group
########################################
Microbial Risk Assessment (MRA) uses large and complex computer models of
food production systems and import or export processes as the basis for
governmental and policy-related decision making. Fundamental to their
applicability is the accurate representation of uncertainty in its many
different forms (for example, uncertainty about input parameters, and
uncertainty about processes (and hence models)). The importance of
separating uncertainty and variability is recognised, but these features are
sometimes incorporated in a relatively ad-hoc, rather than a formal
statistical or probabilistic, manner. The predominant focus to date has
additionally been on parameter uncertainty and methods for appropriate
reflection of this aspect, rather than wider types of uncertainty.
Bayesian methods have several features which are attractive from the point
of view of MRA: first, they describe uncertainty in a formal probabilistic
framework. Secondly, they allow the inclusion of prior belief via a proper
probability ("prior") distribution. Thirdly they provide a natural means of
evidence synthesis, and finally they treat uncertainty and variability in a
properly structured framework. A number of recent papers have begun to
explore the interface between MRA and Bayesian statistics. The potential
offered by this interface is the focus of the presentation.
Through the Environment and Human Health initiative which is jointly funded
by NERC, EA, Defra, the MOD, MRC, The Wellcome Trust, ESRC, BBSRC, EPSRC and
HPA, we established a working group to review the application of Bayesian
approaches in MRA. Our group includes academic and government
representatives from risk analysis, statistics, food safety, mathematical
modelling, veterinary science and public health. To illustrate the breadth
of potential offered by novel methods, we describe a number of case studies
in the context of a model estimating the exposure of humans to VTEC O157
from milk sold as pasteurised, which use the BACCO methodologies (Bayesian
sensitivity analysis, model calibration, model uncertainty and elicitation)
to explore aspects of uncertainties in mathematical and simulation models.
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|