BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:Statistics
SUMMARY:Interim Monitoring of Clinical Trials: Decision Th
eory\, Dynamic Programming and Optimal Stopping -
Chris Jennison\, University of Bath
DTSTART;TZID=Europe/London:20121116T160000
DTEND;TZID=Europe/London:20121116T170000
UID:TALK40392AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/40392
DESCRIPTION:It is standard practice to monitor clinical trials
with a view to\nstopping early if results are suf
ficiently positive\, or negative\,\nat an interim
stage. We shall explain how properties of stopping
\nboundaries can be calculated and how boundaries
can be optimised\nto minimise expected sample size
while controlling type I and II\nerror probabilit
ies.\n\nConstraints on error probabilities complic
ate this optimisation\nproblem. However\, a soluti
on is possible through consideration of\nunconstra
ined Bayes decision problems which are convenientl
y solved\nby dynamic programming. This conversion
to an unconstrained problem\nis equivalent to usin
g Lagrange multipliers. We shall present details\n
of numerical computation for group sequential test
s and their\noptimisation for particular criteria.
We shall discuss a variety of\napplications in cl
inical trial design including the derivation of\no
ptimal adaptive designs in which future group size
s are allowed to\ndepend on previously observed re
sponses\; designs which test both for\nsuperiority
and non-inferiority\; and group sequential tests
which\nallow for a delay between treatment and res
ponse.\n\nSince optimality in the unconstrained pr
oblem can be expressed as a\nsample path property\
, this is an "optimal stopping" problem in the\nla
nguage of probability theory. The computational me
thods we describe\nare\, therefore\, applicable to
such problems and\, in particular\, to\noptimal s
topping problems arising in financial mathematics.
LOCATION:MR12\, CMS\, Wilberforce Road\, Cambridge\, CB3 0W
B
CONTACT:Richard Samworth
END:VEVENT
END:VCALENDAR