ARTICLE IN PRESS
Journal of Biomechanics 40 (2007) 3563–3569
www.elsevier.com/locate/jbiomech
www.JBiomech.com
The impact of adding trunk motion to the interpretation of the role of
joint moments during normal walking
Mausam Patela,, Mukul Talatyb,1, Sylvia Õunpuua,2
a
Center for Motion Analysis, Connecticut Children’s Medical Center, 282 Washington Street, Hartford, CT 06106, USA
b
Gait and Motion Analysis, MossRehab, 60 East Township Line Road, Elkins Park, PA 19027, USA
Accepted 2 June 2007
Abstract
Biomechanical model assumptions affect the interpretation of the role of the muscle or joint moments to the segmental power
estimated by induced acceleration analysis (IAA). We evaluated the effect of modeling the pelvis and trunk segments as two separate
segments (8 SM) versus as a single segment (7 SM) on the segmental power, support of the body, knee and hip extension acceleration
produced by the joint moments during the stance phase of normal walking. Significant differences were observed in the contribution of
the stance hip abductor and extensor moments to support, ipsilateral knee and hip acceleration, and ipsilateral thigh and upper body
power. The primary finding was that the role of the stance hip moment in generating ipsilateral thigh and upper body power differed
based on degrees of freedom in the model. Secondarily, the magnitude of contributions also differed. For example, the hip abductor and
extensor moments showed greater contribution to support, hip and knee acceleration in the 8 SM. IAA and segment power analysis are
sensitive to the degrees of freedom between the pelvis and trunk. There is currently no gold standard by which to evaluate the accuracy of
IAA predictions. However, modeling the pelvis and trunk as separate segments is closer to the anatomical architecture of the body. An 8
SM appears to be more appropriate for estimating the role of joint moments, particularly to motion of more proximal segments during
normal walking.
r 2007 Elsevier Ltd. All rights reserved.
Keywords: Induced acceleration; Segmental power; Trunk motion; Normal gait; Joint moments
1. Introduction
Recent modeling and simulation studies have used basic
principles of linked segment dynamics to understand how a
muscle force or joint moment contributes to the motion of
segments and joints of the entire body (Anderson and
Pandy, 2003; Arnold et al., 2005; Kepple et al., 1997; Liu
et al., 2006; Neptune et al., 2001, 2004; Riley et al.,
2001a, b). This analysis is called induced acceleration
analysis (IAA). This approach has also been expanded to
segmental power analysis to study mechanical energy flow
(Fregly and Zajac, 1996; Neptune et al., 2001, 2004; Seigel
Corresponding author. Tel.: +1 860 380 0164; fax: +1 860 545 8842.
E-mail addresses: Mvpatel@ccmckids.org (M. Patel),
mctalaty@einstein.edu (M. Talaty), Sounpuu@ccmckids.org (S. Õunpuu).
1
Tel.: +1 215 663 6682.
2
Tel.: +1 860 545 8710.
0021-9290/$ - see front matter r 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jbiomech.2007.06.031
et al., 2004). Varying degrees of complexity exist in the
models used in the above studies, depending on the
objective of the study and feasibility of incorporating
the complexity in the model.
The extent to which biomechanical model assumptions
affect the interpretations of muscle function is not fully
clear. Some researchers have used three-dimensional
models in their studies (Anderson and Pandy, 2003; Arnold
et al., 2005; Kepple et al., 1997; Liu et al., 2006; Riley et al.,
2001b; Siegel et al., 2001), whereas others have used planar
models (Fregly and Zajac, 1996; Higginson et al., 2006;
Neptune et al., 2001, 2004). The types of foot–floor
interactions used in the above studies also vary. All the
studies except the one using Anderson’s model (Anderson
and Pandy, 2003) have modeled everything above the hip
joints as a single segment. Some of the results from the
above studies have differed depending on the model used
(Anderson and Pandy, 2003; Kepple et al., 1997; Neptune
ARTICLE IN PRESS
3564
M. Patel et al. / Journal of Biomechanics 40 (2007) 3563–3569
et al., 2004; Riley et al., 2001b). Though some of the recent
studies have started using Anderson’s 10-segment model,
where the pelvis and trunk are modeled as separate
segments, to our knowledge there exists no literature that
documents the impact on the IAA results of modeling the
pelvis and trunk segments as two separate segments versus
as a single segment.
The effect of assuming the pelvis and trunk as a single
segment on the IAA results has not been reported, despite
clear relative motion between the trunk and pelvis during
walking. (Riley et al., 2001c). The combined pelvis–trunk
segment assumption may especially not hold for pathologic
gait, as observed in patients with myelomeningocele who
exhibit excessive trunk motion relative to the pelvis
(Ounpuu et al., 2000). Before performing IAA on the gait
of the patients with myelomeningocele, we aimed to clarify
how upper body modeling assumptions affected the
interpretation of the role of muscles during walking. We
studied segmental power, knee and hip extension acceleration and support of the body (total body center of mass
vertical acceleration) produced by all the joint moments to
infer muscle function.
2. Methods
Motion data from five healthy subjects (65710 kg; 1.6570.11 m; 2371
year; two males and three females) with no known pathology were
collected at Connecticut Children’s Medical Center during normal
walking. Informed consent was obtained from the subjects participating
in the data collection. The subjects were instrumented with 16 reflective
markers on the lower extremities, pelvis and trunk. The subjects were
asked to walk barefoot at a self-selected pace along a pathway on which
three force plates (AMTI, Model # OR6, Watertown, MA) were mounted
in series. Data were collected from each subject using the 512 Vicon
Motion Systems (Vicon Motion Systems, Lake Forest, CA). The testing
protocol for these subjects has been described elsewhere (Davis et al.,
1991). A single trial for each subject was chosen, in which three
consecutive steps contacted the three force plates. Joint moments and
powers were computed using standard inverse dynamics approach (Davis
et al., 1991) using anthropometric data from Dumas et al. (2007). The load
summation for the most proximal segment computed using inverse
dynamics might not equal the product of mass/inertia and measured
acceleration. Skin movement artifact, error in the estimated segment
anthropometry, error in locating the joint center location, etc. are possible
reasons (Plamondon et al., 1996). Loads that make the equations of
motion of the most proximal segment dynamically consistent were
calculated for the upper body and trunk in the 7- and 8-segment models,
respectively, and considered when calculating the model output sum.
The three-dimensional biomechanical model was created in ADAMS
(MSC Software, Ann Arbor, MI) for each subject. The basis of the
biomechanical model used in this work was developed based on the work
done by M. Patel during the research component of her masters work that
was carried out at the Gait and Motion Analysis Laboratory at Moss
Rehab. Two different three-dimensional (3D) models were created for
each subject, a 7-segment (7 SM), 22 degrees-of-freedom (DOF) [Upper
body segment (combined head, arms, torso and pelvis) and bilateral thigh,
shank, feet] model and an 8-segment (8 SM), 25 DOF [trunk (combined
head, arms and torso), pelvis, and bilateral thigh, shank, feet] model. For
both the models, the ankle was modeled as a universal joint, and the hip
and knee as spherical joints. The knee was modeled as a spherical joint for
upcoming analysis of patients with myelomeningocele who may exhibit
some motion in all three planes at the knee (Ounpuu et al., 2000). For the
8 SM the lumbar joint (joint between the pelvis and trunk) was modeled as
a spherical joint. The motion and orientation of the upper body segment in
the 7 SM was tracked using the pelvis markers but the anthropometry of
the combined pelvis and trunk was used, similar to Kepple et al. (1997).
Foot–floor interaction was modeled as a revolute joint at the measured
center of pressure (CoP) obtained from the forceplates during the entire
stance phase.
The segmental power, knee and hip extension acceleration and support
of the whole body produced by the joint moments and gravity during the
stance phase were analyzed for both the 7 and 8 SMs. The total body
center of mass (COG) vertical acceleration was assumed to be
representative of the support of the body. Input to the model was body
configuration (position and orientation), location of the center-ofpressure, computed joint moments and the offset loads. The model was
reconfigured for each analysis frame using motion capture data. Each joint
moment’s contribution to segmental power, COG vertical acceleration
and joint angular accelerations was calculated for each frame by the solver
module in ADAMS, using methods similar to ones previously described
(Kepple et al., 1997; Seigel et al., 2004). For the 8 SM, the lumbar joint
moments were also applied. For ease of comparison of the upper body
power for the two models the power of the pelvis and trunk segments were
summed to estimate the contribution of each joint moment to upper body
power for the 8 SM. This was then compared to the upper body power of
the 7 SM. The segmental power results were normalized to body mass to
compare the data across subjects.
The model output sum was compared with the results computed from
the motion capture data to validate the model (Seigel et al., 2004). The
term ‘reconstruction’ was used to describe the computed model output
sum. The contribution of the coriolis and centripetal forces during normal
walking was assumed to be negligible.
3. Results
Induced acceleration and segmental power using the two
different models were computed for the five subjects. The
results for a representative subject are graphed in Figs. 1–3,
5 and 7; and the averages of the five subjects are graphed in
Figs. 4, 6 and 8.
The inverse dynamics solution was used to verify the
model (Kepple et al., 1997). The data computed from the
model output were reasonable compared to the motion
data (Fig. 1). Similar reconstructions were obtained for the
two models (Fig. 1).
3.1. Segmental power
Key differences in hip contribution to upper body power
between the 7 and 8 SMs were noted (Fig. 2). The stance
hip moment added energy to the upper body in early stance
and removed energy from the upper body during late
stance for the 7 SM, whereas the opposite was observed for
the 8 SM. The magnitude of the power delivered or
removed from the thigh, upper body and contra lateral leg
by the hip moment was considerably greater for the 8 SM.
The ankle moment removed less power from the upper
body segment and for a shorter duration during early
stance for the 8 SM.
3.2. Support of the body
The contribution of the joint moments to the COG vertical
acceleration for the 7 and 8 SMs are shown in Figs. 3
ARTICLE IN PRESS
M. Patel et al. / Journal of Biomechanics 40 (2007) 3563–3569
3565
Fig. 1. Net stance leg joint power, body center of gravity vertical acceleration, hip and knee extension acceleration as computed from the model data and
the motion data for a single representative subject during the stance phase of gait. The solid line is as calculated/measured from the motion data, the dotted
line is the 7-segment model reconstruction and the dash–dot line is the 8-segment model reconstruction.
Fig. 2. Segmental power due to the stance leg hip, knee and ankle moment during the stance phase of gait for a representative subject. The abbreviations
used in the legend are explained here. ITh: ipsilateral thigh (dash-dot), ISk: ipsilateral shank (dash-dash); IFt: ipsilateral foot (solid); Contra: contra lateral
leg (dot); and UB: upper body (dotted line). Note: the upper body power for the 8-segment model was the sum of contributions to the pelvis and trunk
segments.
ARTICLE IN PRESS
3566
M. Patel et al. / Journal of Biomechanics 40 (2007) 3563–3569
Fig. 3. Contribution of the joint moments and gravity to body center of gravity vertical acceleration (support) during the stance phase of gait for a
representative subject. The abbreviations used in the legends are explained here. SIA: sagittal ipsilateral ankle moment, CIA: coronal ipsilateral ankle
moment, SIK: sagittal ipsilateral knee moment; CIK: coronal ipsilateral knee moment; SIH: sagittal ipsilateral hip moment; CIH: coronal ipsilateral hip
moment; Contr: contra lateral leg moments; Grav: passive resistance of bones and joints to gravity; SL: sagittal lumbar moment; CL: coronal lumbar
moment.
and 4. For the 8 SM, the hip extensors during early stance and
hip abductors during mid-stance provided greater support to
the body than that observed in the 7 SM (Fig. 4). Also lumbar
sagittal and coronal moments provided support to the body
during the second half of stance for the 8 SM (Fig. 3).
3.3. Knee and hip extension acceleration
The differences observed in the distribution of the
contributions to the knee angular acceleration between the
7 and 8 SMs were small (Figs. 5 and 6), whereas
considerable differences were observed in the magnitude of
the contributions for the hip angular acceleration (Figs. 7
and 8). The 7 SM showed prolonged contribution of ankle
sagittal moment to flexion acceleration of the knee in early
stance. The contributions of the ipsilateral hip extensors and
abductors to hip and knee acceleration during stance were
greater for the 8 SM than for the 7 SM (Figs. 5–8). For the 8
SM, hip abductors caused knee flexion acceleration during
first half of stance and hip extension acceleration for the
second half of stance. However, for the 7 SM they had
minimal contribution for those phases (Fig. 6).
4. Discussion
We aimed to evaluate the effect of modeling the pelvis and
trunk as a single versus separate segments on acceleration
Fig. 4. Mean contribution across all five subjects of the major
contributors to the body center of gravity vertical acceleration for the
7- and 8-segment models during early stance (up to 20% of the gait cycle)
unless specified.
analysis calculations during the stance phase of normal
walking. Specifically, we evaluated segmental power, support of the body, and knee and hip extension acceleration
produced by the joint moments for the two models.
Differences were observed predominantly in the contribution of the ipsilateral hip sagittal and coronal moments to
support, upper body and ipsilateral thigh power, and hip
and knee acceleration. The polarities of the contributions,
which are indicative of the role of the muscle or joint
ARTICLE IN PRESS
M. Patel et al. / Journal of Biomechanics 40 (2007) 3563–3569
3567
Fig. 5. Contribution of the joint moments and gravity to knee extension acceleration during the stance phase of gait for a representative subject. The
abbreviations used in the legends are explained here. SIA: sagittal ipsilateral ankle moment, CIA: coronal ipsilateral ankle moment, SIK: sagittal
ipsilateral knee moment; CIK: coronal ipsilateral knee moment; SIH: sagittal ipsilateral hip moment; CIH: coronal ipsilateral hip moment; Contr: contra
lateral leg moments; Grav: passive resistance of bones and joints to gravity.
Fig. 6. Mean contribution across all five subjects of the joint moments to
knee extension acceleration during early stance (5–25% of the gait cycle).
moment, were the main focus. Differences in the magnitudes
of contributions were also noted, but their importance at
this time is less clear.
It is important to consider some modeling assumptions
used in the present study before discussing the results.
First, the same CoP location that calculated from actual
force plate data was used for both the 7 and 8 SMs.
Changing the upper body orientation, as done in the 7 SM,
would likely change the CoP, which would affect the
computed lower extremity kinetics (Westwell et al., 2005).
We did not recompute the CoP, as the focus of the present
work was to illustrate the impact of modeling the trunk
and pelvis as separate versus a single segment on the IAA
results using biomechanical models similar to those used in
the literature (Seigel et al., 2004; Anderson and Pandy,
2003). Secondly, the foot–floor interaction was assumed as
a rigid contact of a revolute joint at the CoP during the
entire stance phase. This assumption may not hold for the
entire stance phase, which may be responsible for some of
the differences between the reconstruction and motion data
as seen in Fig. 1. Lastly, a spherical knee joint was used in
the current study instead of the most commonly used hinge
joint (Seigel et al., 2004). The impact of using a spherical
joint instead of a hinge joint on the differences observed
between the 7- and 8-segment models was tested. It was
found that the IAA results were not sensitive to the two
extra knee degrees of freedom. The differences between the
7- and 8-segment models remained about the same, with
small changes in the magnitude of the contributions
observed for some of the variables for both the 7- and
8-segment models.
Despite the similarity of model output reconstructions
(Fig. 1), the distributions of the joint moments, particularly
those of the hip to the segmental power or accelerations
were appreciably different (see Figs. 2–8). Chen (2006) had
found that increasing the degrees of freedom of the model
he used resulted in greater power redistribution attributed
to the joint moments. For our study, this was not true for
all the joint moments. Extra lumbar degrees of freedom
increased the contribution of the ipsilateral hip moments to
ARTICLE IN PRESS
3568
M. Patel et al. / Journal of Biomechanics 40 (2007) 3563–3569
Fig. 7. Contribution of the joint moments and gravity to hip extension acceleration during the stance phase of gait for a representative subject. The
abbreviations used in the legends are explained here. SIA: sagittal ipsilateral ankle moment, CIA: coronal ipsilateral ankle moment, SIK: sagittal
ipsilateral knee moment; CIK: coronal ipsilateral knee moment; SIH: sagittal ipsilateral hip moment; CIH: coronal ipsilateral hip moment; Contr: contra
lateral leg moments; Grav: passive resistance of bones and joints to gravity.
Fig. 8. Mean contribution across all five subjects of the joint moments to
hip extension acceleration during early stance (5–25% of the gait cycle)
unless specified.
the segmental power, but the contribution of the ipsilateral
knee moment to the segmental power did not change much
and the contribution of the ipsilateral ankle moment to the
segmental power decreased (Fig. 2). The joint reaction
forces more proximal to the body and on the contra-lateral
side were the most affected when modeling a separate trunk
and pelvis segment.
The main reason for the observed differences between
the two models was the extra degrees of freedom allowed
for the trunk in the 8 SM and not differences in trunk
orientation or its center of mass location. Because the 7 SM
used pelvis markers to position and orient (POSE) the
trunk, the trunk center of mass location and inertia
orientation were different between the 7 and 8 SMs—in
addition to the extra degrees of freedom allowed in the
latter. Either could have caused the significant differences
we observed. To clarify further, a test was performed where
the trunk POSE was forced to be equal to the pelvis POSE
for the 8 SM, while keeping the joint between the pelvis and
trunk as a spherical joint. This was named the 8-segment
dummy. The 8-segment dummy energy flow results were
similar to the 8 SM with subtle differences in the magnitude
of the contributions, indicating that most of the differences
between the 7 and 8 SMs were due to the extra degrees of
freedom allowed for the trunk in the 8 SM.
The segmental power attributed to the joint moments for
the 7 SM in our study was generally consistent with that
reported in the literature (Neptune et al., 2004; Seigel et al.,
2004), using 7 SMs. Differences in the segmental power
between the 7 and 8 SMs were predominantly observed in
the contribution of the hip moments. The hip flexor
contribution to the upper body power changed on
modeling the pelvis and trunk as separate segments.
Neptune et al. (2004) and Seigel et al. (2004) reported that
the hip flexors redistributed the power from the upper body
to the ipsilateral leg during pre-swing, as we did in the 7
SM. But for the 8 SM, the opposite was observed. Transfer
of power into the upper body, during pre-swing, is in
ARTICLE IN PRESS
M. Patel et al. / Journal of Biomechanics 40 (2007) 3563–3569
agreement with suggestions of others that the hip flexors
assist forward acceleration (Sadeghi et al., 2001).
The role of the ipsilateral hip extensors during early
stance and ipsilateral hip abductors during mid-stance to
support appears to be underestimated if the pelvis and
trunk segments are fused as a single segment (Figs. 3 and
4). Unlike the 7 SM, for the 8 SM the hip abductors
had significant contribution to support during mid-stance
(Fig. 3) for most of the subjects; similar to that observed by
others (Anderson and Pandy, 2003; Liu et al., 2006), who
modeled the pelvis and trunk as separate segments.
Substantially greater hip abductor contribution was
observed for two of the five subjects using the 8 SM, who
had their trunk leaning more towards their contra lateral
side. For the 7 SM, the support provided by the hip
abductors throughout stance remained small for all the five
subjects; similar to that observed by Kepple et al. (1997),
who used a 7 SM. It appears that to quantify the role of the
hip abductors to support, it is essential to model the pelvis
and trunk as separate segments and allow the three
rotational degrees of freedom between the two segments.
The overall trend of the major contributions other than
by the ipsilateral hip abductors to hip and knee accelerations did not differ much for the two models. However,
some discrepancies were observed in the magnitude of the
contributions. Hip and knee extensors accelerated the hip
and knee joints into extension during the first half of stance
for both the models, similar to that observed by Neptune et
al. (2004) and Arnold et al. (2005), but greater hip extensor
contribution was observed for the 8 SM. Unlike the 7 SM,
the hip abductors in the 8 SM were found to cause knee
flexion acceleration during first half of single support phase
(Fig. 6), and hip extension acceleration during the second
half of stance (Fig. 8), similar to Arnold et al. (2005), who
modeled the pelvis and trunk as separate segments.
This study documents the segmental power flow due to
joint moments using a model which models the pelvis and
trunk as separate segments, which to date does not exist in
the literature. From the current study it was found that
modeling the pelvis and trunk as separate segments impacts
the interpretation of the role of the joint moment during
normal walking. Differences in the trends of the contribution of the stance hip moment to the ipsilateral thigh and
upper body power, support, hip and knee sagittal acceleration were observed between the 7 and 8 SMs. This
changes our understanding of the possible role of the hip
moment during stance phase. In the future, we plan to use
the 8 SM to analyze the mechanisms used to walk by
patients with myelomeningocele and study the role played
by the lumbar moments to substitute for the lower
extremity muscle weaknesses.
Conflict of interest
None of the authors of this manuscript (Mausam
Patel, Dr. Mukul Talaty and Sylvia Õunpuu) have any
financial or personal relationships with other people or
3569
organizations that could inappropriately influence (bias)
this work. Also the material in the manuscript has not been
and will not be submitted for publication elsewhere except
as an abstract.
Acknowledgments
Some of the computational aspects of this work were
performed at MossRehab. The authors gratefully acknowledge the comments on an earlier draft of this manuscript by
Alberto Esquenazi, MD.
References
Anderson, F., Pandy, M., 2003. Individual muscle contributions to
support in normal walking. Gait & Posture 17, 159–169.
Arnold, A., Anderson, F., Pandy, M., Delp, S., 2005. Muscular
contributions to hip and knee extension during single limb stance
phase of normal gait: a framework for investigating the causes of
crouch gait. Journal of Biomechanics 38, 2181–2189.
Chen, G., 2006. Induced acceleration contributions to locomotor
dynamics are not physically well defined. Gait & Posture 23, 37–44.
Davis, R., Ounpuu, S., Tyburski, D., Gage, J., 1991. A gait analysis data
collection and reduction technique. Human Movement Science 10,
575–587.
Dumas, R., Cheze, L., Verriest, J., 2007. Adjustments to McConville et al.
and Young et al. body segment inertial parameters. Journal of
Biomechanics 40, 543–553.
Fregly, B., Zajac, F., 1996. A state-space analysis of mechanical energy
generation, absorption and transfer during pedaling. Journal of
Biomechanics 29, 81–90.
Kepple, T., Siegel, K., Stanhope, S., 1997. Relative contributions of the
lower extremity joint moments to forward progression and support
during gait. Gait & Posture 6, 1–8.
Liu, M., Anderson, F., Pandy, M., Delp, S., 2006. Muscles that support
the body also modulate forward progression during walking. Journal
of Biomechanics 39, 2623–2630.
Neptune, R., Kautz, S., Zajac, F., 2001. Contributions of the individual
ankle plantarflexors to support, forward progression and swing
initiation during walking. Journal of Biomechanics 34, 1387–1398.
Neptune, R., Zajac, F., Kautz, S., 2004. Muscle force redistributes
segmental power for body progression during walking. Gait & Posture
19, 194–205.
Ounpuu, S., Thomson, J., Davis, R., DeLuca, P., 2000. An examination of
the knee function during gait in children with myelomeningocele.
Journal of Pediatric Orthopaedics 20, 629–635.
Plamondon, A., Gagnon, M., Desjardins, P., 1996. Validation of two 3-D
segment models to calculate the net reaction forces and moment at the
L5/S1 joint in lifting. Clinical Biomechanics 11, 101–110.
Riley, P., Della, C., Kerrigan, D., 2001a. Effect of age on lower extremity
joint moment contributions to gait speed. Journal of Biomechanics 14,
264–270.
Riley, P., Della, C., Kerrigan, D., 2001b. Propulsive adaptation to
changing gait speed. Journal of Biomechanics 34, 197–202.
Riley, P., Della, C., Kerrigan, D., 2001c. Comments on propulsive
adaptation to changing gait speed. Journal of Biomechanics 34,
1669–1670.
Sadeghi, H., Sadeghi, S., Prince, F., Allard, P., Labelle, H., Vaughan, C.,
2001. Functional roles of ankle and hip sagittal muscle moments in
able-bodied gait. Clinical Biomechanics 16, 688–695.
Seigel, K., Kepple, T., Stanhope, S., 2004. Joint moment control of
mechanical energy flow during normal gait. Gait & Posture 19, 69–75.
Westwell, M., Ounpuu, S., Bell, K., 2005. Effect of upper body motion on
lower extremity kinetics. In: Proceedings of the 10th Gait and Clinical
Movement Analysis Annual Meeting, Portland, OR.