.Note slight change of venue - 4 minutes (max.) walk
. from MacKay Building
CENTRE FOR MEDICAL STATISTICS
SEMINAR SERIES on BIOSTATISTICS 1999/2000
NO 8: Wednesday, June 14th, 2000, at 2:30 pm
Modelling prognostic factors in primary breast cancer
by
Professor Patrick Royston
(MRC Clinical Trials Unit, London)
There is considerable interest in the oncological literature
in novel factors which may predict the clinical course of
patients with primary breast cancer. At least 160 such
factors have been claimed to have prognostic value, but
few have been independently validated. One problem is
the lack of an accepted multivariable ‘standard’ model
against which new factors may be assessed. I shall
describe one approach to deriving such models, based
on the use of fairly simple parametric functions
(fractional polynomials) which keep continuous factors
continuous. Arbitrary categorisation of factors is thus
avoided. I shall discuss the assessment of non-linearity
and goodness of fit, including the use of a method based
on cubic smoothing splines to see if the parametric
model has overlooked some important aspect of the
data. Depending on time, I will aim to talk about some
of the following issues: defining prognostic groups;
measuring prognostic ability of a model; assessing the
prognostic ability of an additional factor; time-dependent
effects. The data from a large cohort of breast cancer
patients from the Netherlands will be used to illustrate
the methods.
All welcome!!
Venue:
Room CBB0.004
Chancellor's Building - Extension
(Bananna Wing)
Keele University
http://www.keele.ac.uk
http://www.keele.ac.uk/depts/ma/seminars/medstats.html
http://www.keele.ac.uk/university/campus/maps/camus.htm
_____________________________
--
_____________________________
Prof. Gilbert MacKenzie
Director,
Centre for Medical Statistics
Keele University
Staffordshire ST5 5BG
England
Tel: (01782) 583651
Fax: (01782) 584268
______________________________
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|