Back to the Top
The responsible use of models.Or how to achieve a semblance of validity.
The user of any tool or technique is responsible to see that the results
obtained are satisfactory for the intended use. How can this be done in
the face of uncertainty, unknown answers, complex and sometimes obscure
algorithms, and calculations which cannot be checked. Clearly these are
not tasks to be undertaken by an individual who wishes to be productive
in their work or research. There are many guides to creating code with
minimal errors. In the general field of modeling and simulation there is
a vast literature on verification and validation. Very little of this is
of value to the individual user. However, there is a heuristic approach
which can limit the degree of error that one may make with modeling
programs which attempt to fit data. It is usually possible to simulate a
functional approximation of the problem of interest. In the simulation
one specifies (and therefore knows) the values of all parameters used.
One can then run the model fitting program and see how close its values
come to the actual known values. If they are close then it lends
confidence to the results obtained by running the program on the actual
data.
Meyer Katzper, Ph.D.
FDA/CDER/ODEV/DAAODP
katzper.-a-.cder.fda.gov
(301) 827-2563
PharmPK Discussion List Archive Index page
Copyright 1995-2010 David W. A. Bourne (david@boomer.org)