>Late Reminder
CENTRE FOR MEDICAL STATISTICS
SEMINAR SERIES on BIOSTATISTICS 1998/99
NO 8 Wednesday, March 31st, 1999 2:30pm
"Bernoulli Bandits and Clinical Trials"
Professor Peter Jones
(Keele University)
Several modifications of the Bernoulli two armed bandit with finite
horizon are discussed. A Bayesian approach is adopted and the
designs are evaluated by using dynamic programming and backward
induction. In addition to designs which maximise expected return,
those which minimise the expected number on the poorer arm will
be discussed. A simple terminal decision rule is introduced and the
probability of correct selection is obtained. The effect of early
stopping
is considered and expected sample sizes are calculated for the optimal
procedures. The effect of group sequential sampling is evaluated and
some sub-optimal schemes are suggested. Applications to the planning
of clinical trials are noted.
Room M222
The MacKay Building
Keele University
http:/www.keele.ac.uk/
All Welcome !
Gilbert
--
_____________________________
Prof. Gilbert MacKenzie
Director,
Centre for Medical Statistics
Keele University
Staffordshire ST5 5BG
England
Tel: (01782) 583651
Fax: (01782) 584268
______________________________
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|