presents
Mendel Fygenson
Pessimistic Modeling and Inference in
the Risk Evaluation of Carcinogens
ABSTRACT
To set standards for exposure to a carcinogen,
regulatory agencies employ the benchmark dose (BMD) method to determine the
“virtually safe dose” (VSD). They claim
that the BMD method is inherently conservative. In this talk I address the question of just
how conservative it is. The question is
particularly crucial in the risk evaluation of carcinogens that have direct or
indirect benefits. In such cases, there are real costs associated with VSD that
are excessively protective of the public’s health.
I’ll present insights that come from
viewing risk evaluation using the BMD method in the context of modern economic
decision theory. In particular, I’ll
introduce two non-parametric families of pessimistic (optimistic) distributions
and show how these provide for a characterization of the current BMD method and
for a semi-parametric lower bound on the risk associated with exposure to the
VSD. I’ll illustrate these results using
data used by the EPA in evaluating the carcinogenicity of chronic exposure to
bromate in drinking water.
Friday, April 20, 2007
3:35 - 4:35 pm
206 Cox Hall
Refreshments will be served on the second floor of Dabney Hall (left of Room 222) at 3:00 pm