View Full Version : Summary of normalizing gait

David Davis
02-17-1997, 10:07 PM
Dear All,

I recently wrote the following question:

"I hope that someone may be able to help / put me straight. I am currently
working on gait analysis during load carraige. This involves subjects of
differing sizes , with differing loads free speed walking over a force plate
and recording the results. Now in order to accurately compare the results I
at first thought that expressing the forces as a percentage of body weight /
total weight was best. But it occurred to me that with free speed walking,
intrasubject variation of velocities as well as intersubject variation of
velocities occurred and (this is the bit that concerns me) since velocity is
a factor of acceleration and acceleration is a factor of force, are not the
ground reaction forces affected by velocity changes. Or do we safely assume
that during terminal swing the only acceleration acting is gravity??
I hope someone can explain a method of normalizing for this or at least tell
me of an experimental protocol that avoids this or tell me I'm wrong!!!! , I
will post a summary of replies."

I'd like to thank all those who replied, the information I have recieved has
been of great use.

Here follows is a summary of the replies I received. At the end of the
document are the references that people supplied/lead me to, for those


Walking velocity affects GRF and in the Anterior-Posterior direction. This
is why we ask the subjects to start and finish at least 7-10m before and
after the force plate, so they maintain their normal speed. We assume
constant velocity, so zero walking acceleration. We normalise our results
with body weight and total stance phase. There is always intra- and inter-
subject variability in GRF (time but also in frequency domain analysis). For
the frequency domain variability check the December 1996 issue of Spine the
work by Giakas and Baltzopoulos.

"Craig Nevin"

The variation in velocity of a subject has vertical and horizontal
components. The vertical component, being subject to gravity, is
independant of mass. The horizontal fluctuation is dealt with by assuming
that the body is moving at a constant veleocity, any acceleration of the
body during one part of the cycle is recovered during another part of the
gait cycle (equal and opposite reactions to maintain the costant speed).
Since the gait cycle is invariably studied at a constant velocity, the
movement is without beginning or end, which is perpetual motion.

Jesus Dapena

On can divide the forces by the body weight. When you divide by body weight,
in essence you are computing accelerations (force/mass) --or more
precisely, values that are directly proportional to accelerations, since
you are dividing by weight and not by mass. You are calculating how much
each horizontal and vertical force contributes by itself to the
acceleration of the whole system (if you are dividing by total system
weight or mass).

Irene S McClay

Most all of the literature reports GRF in bodyweights. However, you are
correct in assuming that gait velocity will influence GRF (Doris Miller has
published in this area). There are two possible ways to deal with this.
Either you control the speed or you can treat velocity as a covariate as you
analyze the data.

"D. Gordon E. Robertson"

In a previous study we compared the subjects against their own unloaded
walking patterns. For example, the mediolateral ground reaction forces were
averaged for unloaded walking and a mean and a standard deviation curve were
produced. The same was done for a loaded condition. We would say the the
loaded condition was significantly different from the unloaded condition
whenever it fell outside the mean +/- a confidence interval of the unloaded
condition. The confidence interval was 1.96*s.d. which represents the 95%
confidence level.

Scott & Joanne Aldworth

At the BloorView Medical Centre in Toronto, Ontario, Canada whilst
performing a gait analysis investigation between normals, above and below
knee amputees in children based on age differences. We normalized to the
weight of the subject and it seems that we only used gravtity at the point
of heel strike.

Sylvain Grenier

We normalize work to velocity * mass, in "free speed" walking trials over a
force platform.

"Vladimir M. Zatsiorsky"

Try the Local Proportional Scaling method (see the Motor Control, v. 1, #1,

"Henk C. Schamhardt"

It may be better to collect baseline data of each subject, and compare
loaded data with those. What I mean is a certain kind of paired comparison.
Dividing GRF by body mass (whether or not including additional mass)
neglects the influence of speed on GRF. Since a tendency towards lower
speeds at increasing loads may be expected, this might seriously interfere
with your data.
The influence of walking speed on GRF has to be dealt with separately (one
should avoid trying to get subjects to walk at a forced speed) for example
by application of appropriate statistical techniques such as ANOVA.

"Ruben Lafuente Jorge"

Velocity is a main factor affecting gait curve morphology, but not gait
performance. Gait performance is the same irrespective of speedSome other
factors affecting gait performance and also curve morphology are pathology
(of course), type of shoe, sex and age. There is at present no scheme for
speed normalization. Maybe you could think about the inverse pendulum model
and try to normalise the centrifugal force or consider velocity an explicit
factor which must be taken into consideration for determining the reference
curves. That is, you calculate separately average curves for slow speeds,
for medium speeds and for fast speeds. This way, you can get rid of a part
of the variability.

************************************************** **************************

Giakas G., Baltzopoulos ?., December 1996, ...frequency domain analysis...,
Spine (incomplete ref.)

???, 1997, Local Proportional Scaling Method, Motor Control, Vol.1,No.1,
pp.?? (incomplete ref.)

Tatara T., Iida T., Walking patterns of normal human gaits with loads on a
level, Biomechanics X-A, pp.397-401, Jonsson B.(ed.), Human Kinetics:
Champaign Il,1987

Millar J.F., Stamford B.A.,1987, Intensity and energy cost of weighted
walking vs running for men and women, Journal of applied physiology, vol.64,

Martin P.E., Heise G.D., Morgan D.W., 1993, Interrelationships between
mechanical power, energy transfers, and walking and running economy.
Medicine and science in sports and exercise, Vol.25, pp.508-515

Webb P et al., 1988, The work of walking: a calorimetric study, Medicine
and science in sports and exercise, Vol.20, pp.331-337

Nagle F.J., Webb, Wanta., 1990, ???, Medicine and science in sports and
exercise, Vol.22, pp.540-544

Sun M., Hill J.O. , 1993, A method for measuring mechanical work and
efficiency during human activities. Journal of biomechanics, Vol.26, pp.229-

Crowe A et al., 1996, The influence of walking speed on parameters of gait
symmetry determined from ground reaction force, Human movement science,
15(3) pp. 347-367

DeVita, P., April 1994, The selection of a standard convention for analyzing
gait data based on the analysis of relevant biomechanical factors, Journal
of biomechanics, Volume 27, Number 4, pp.501-

Fioretti S, 1996, Signal processing in movement analysis, Human movement
science, 15(3) pp. 389-410

Vaughan CL, 1996, Are joint torques the holy-grail of human gait analysis,
Humna movement science, 15(3) pp.423-443

Department of Human Sciences
Loughborough University
LE11 3TU
Tel: 01509 223086
Fax: 01509 223941
Email: D.Davis@lboro.ac.uk