SYNTHESIS OF HUMAN MOVEMENT—FORWARD SOLUTIONS - Biomechanics and Motor Control of Human Movement - page 202

a surgeon might ask the question, “What would happen if I transferred a
muscle from one insertion point to another, or what would happen if
a muscle tendon were released (lengthened) to reduce specific muscle
spasticity?” Or a coach might ask, “Is the movement pattern of my runner
optimal, and if not, what changes in motor pattern might improve it?” Or a
basic researcher might have a certain theory of the motor control of gait and
might wish to test that theory.
However, before any valid answers are forthcoming, the link-segment rep-
resentation of the anatomy must first be valid. A necessary (but not sufficient)
condition that has been tested by researchers is that of internal validity. Such
a test requires an inverse solution to calculate the moments at each of the
joints in each of the required planes (sagittal, coronal, etc.). Then, by using
these motor patterns as inputs along with measured initial conditions, the
forward solution should reproduce the originally measured kinematics. If the
model does not pass that test, it is fruitless to use the model to answer any
functionally related questions. All that will result will be an erroneous set of
motor patterns overlaid on an erroneous biomechanical model.

8.1 REVIEW OF FORWARD SOLUTION MODELS

Human locomotion is the movement that has attracted the most attention with
researchers. Because of the complexity of the movement and the link-segment
model, certain oversimplifications were made or the simulation was con-
fined to short periods of the movements. Townsend and Seireg (1972) mod-
eled the human with massless rigid lower limbs with 1 degree of freedom
at each hip (flexion/extension). Hemami (1980) proposed a three-segment
three-dimensional model with rigid legs and no feet, and Pandy and Berme
(1988) simulated single support only, using a five-segment planar model with
no feet. Obviously, such serious simplifications would not produce valid
answers. Even with more complete models, many researchers constrained
parts of their models kinematically (Beckett and Chang, 1968; Chao and
Rim, 1973; Townsend, 1981) by assuming sinusoidal trajectories of the trunk
or pelvic segments. Such constraints violate one of the major requirements
of a true simulation. Initial work in our laboratory (Onyshko and Winter,
1980) modeled the body as a seven-segment system (two feet, two legs, two
thighs, and an HAT segment), but the model did not satisfy the requirements
of internal validity because of certain anatomical constraints (sagittal plane
movement only at all three joints plus a rigid foot segment). The model was
eventually made to walk but only by altering the moment patterns. Such
results should alert researchers that two wrongs can make a right. An incom-
plete model will result in a valid movement only if faulty motor patterns are
used. More complex models (more segments, more degrees of freedom at
each joint) have been introduced (Hemami et al., 1982 b ; Chen et al., 1986),
but as yet no internal validity has been attempted.