Richard Baker
01-08-2002, 11:11 AM
Dear all,

I'd like to agree very much with Chris's comments. From deep within
clinical gait analysis my assumption is that the methods we use are
reasonably accurate for the populations we are examining. However I do not
know. The applied biomechanics community has been negligent to an
extraordinary degree in not specifying the validity of its results. Most
labs produce "normal databases" which give some indication of repeatability
of data (generally taken from a population other than the one they are
working with). These studies however give no information about the validity
of the data. Hidden in the academic literature are a variety of studies
which throw some light on various aspects of our modelling but these
generally stay there. Errors occur, and are unspecified at all levels of
data capture and processing. Try pressing an equipment manufacturer for a
specification of the accuracy of marker detection - and then put it to the
test in a real lab!

To my mind the most pressing need for clinical biomechanics is the
development of measures of validity. Without these we don't know which
data is valid and which is not. Without this information it is also
impossible to know which measurements we can accept and which we to
development further. I am convinced that within the dataset we use some
data are reliable and some are not. I have my instincts as an engineer to
try and determine which are which, but I have very little hard data to
support this. Persuading clinicians to place less emphasis on some data
than on others as a consequence of measurement/modelling methodologies is
virtually impossible in the absence of a clear indication of the validity
of the various measures.

Of course there is a methodological problem here in that to assess the
validity of one model it is generally necessary to have another model that
one believes to be a close match to reality. It may be that Prof Hatze's
call for the development of new generation of models is the precursor to
understanding the existing models. There are however other methods of
assessing validity of existing models, from sensitivity analysis to input
parameters to good old fashioned informed discussion of likely error
sources. Can I suggest that what the clinical biomechanics community
requires is to drag this discussion out of the academic back room and into
the forefront of its considerations.

Of course we have to discuss this in context. The one thing we know less
about than the validity of our data is how to use them once we've got them.
Its easy to get depressed by the limitations of our own techniques but,
with the possible exception of transverse plane kinematics, I'm happy that
the data we supply to clinicians are appropriate for their level of
understanding of how to interpret those data in clinical decision making.

This probably reveals a tension which is present in any applied science.
The most pressing need I see for coming years is not for more complex
models but for the appropriate incorporation into clinical practice of the
models that we already have. I am generally fairIy depressed by the level
of biomechanical understanding I see from both bioengineers and clinical
colleagues at conferences, in peer reviewing publications and in clinical
audit. There is still a huge amount that first generation biomechanics has
to offer the clinical community on top of what it offers at present. I can,
however, quite understand those who see biomechanics as a discipline in its
own right and want to pioneer the second generation. The tension between
pure and applied aspects of any discipline is positive. Let the purists
pioneer the second generation, but lets all concentrate on applying the
first generation appropriately at the same time.

Richard Baker

Richard Baker PhD CEng
Gait Analysis Service Manager and Director of Research
Hugh Williamson Gait Laboratory
Royal Children's Hospital
Flemington Road,
Victoria 3052, Australia

Tel: +61(0)3 9345 5354
Fax: +61(0)3 9345 5447
e-mail: bakerr@cryptic.rch.unimelb.edu.au

To unsubscribe send SIGNOFF BIOMCH-L to LISTSERV@nic.surfnet.nl
For information and archives: http://isb.ri.ccf.org/biomch-l