Transcript 0204

Introduction of PCA and energy
flow pattern in lower limb
Reporter:Yu-shin Chang
Date: 99/02/05
Questions ?

Would healthy elders with decreased ankle joint
power tend to adopt hip strategies when walking
faster?

Weaker plantar flexor muscle induced lesser
ankle joint power generation in elders would
cause energy flow disruption and redistribution
thus disturb propulsion

Would healthy elders reveal distinct power
coordination from healthy young adults?
Objective

To investigate lower limb energy transfer,
changes in joint coordination and compensation
that elders adopt due to aging.
Design
Principal component analysis(PCA)
 Mechanical energy model

Participants

11 healthy elder(mean=68 yrs)
11 healthy young adults(mean=25yrs)

All subjects walked along a 10m walkway in
self-selected speed & faster speed

Choose 3 successful trials to analysis
Main outcome measurements:

Temporal-spatial parameters and kinematic and
kinetic parameters
◦ Joint angles
◦ Joint moment
◦ Joint power
◦ Segmental power
Results:
 Elders:
◦ reduced peak ankle power generation
◦ More knee power absorption
◦ More hip flexor power
weak plantar-flexors in elders for locomotion.
◦ At fast speed generate more hip concentric
power than self-selected speed
hip strategy
Conclusion:

Decreased ankle power in elders induce
compensation of other muscular.
 energy distributing abnormally thus evoke hip flexor
and knee extensor simultaneously act to balance.
Lower thigh segment energy due to insufficient
ankle power and need mor hip flexor power for
larger stride length.
 Produced lesser power from transverse and
frontal plane

more instable in gait.
Data analysis
Kinematic data analysis
Hip, knee, ankle joint angles
 Linear / angular velocity

Kinetic data analysis

COP: calculated from GRF and moment measured on
force plate.

Joint reaction force and moment: use inverse
dynamics

Joint power: product of joint moment and angular
velocity

Segment mechanical energy
◦ Kinetic energy(Ek):
the energy possessed by a body in motion
Ek= 1/2mv2 + 1/2 Iω2
(linear) + (rotation)
◦ Potential energy(Ep):
the energy is acquired through a change in
configuration of body
Ep=mgh
total mechanical energy E = Ek + Ep

Segmental power terms
◦ Segmental joint force power(translation)
Pft = F V
●
+: energy flow “into ”segment
- : energy fow “out ” segment
◦ Segmental joint force power(rotation)
Pfr = ωs (r x F)
◦ Segment joint moment power(Pm)
Pm = M ωs
●
●
+: flow of energy from muscle into segment
- : flow of energy into muscle
Segmental total power(Ps)
Ps = (Pftp+ Pftd)+(Pfrp + Pfrd ) +(Pmp + Pmd )
Segmental power terms

Power flow pattern model

the lower extremity
was treated as 4
segment linked system.
◦
◦
◦
◦
Pelvis
Thigh
Shank
Foot
Principal component analysis (PCA)
PCA model

PCA model(Principal Components Analysis):
◦ reduce data dimensionality by performing a
covariance analysis between variables and
expressiong the data in such a way to highlight
their similarities and differences.
◦ Find the direction in the data with the most
variation

Use 2 PCA models to identify
◦ joint coordination by using power flow terms
that across joints and segments
◦ The most important factors that could
discriminate gait pattern of healthy elders from
healthy young

PCA model 1
◦ Step1: calculate the covariance matirx of data
parameters
◦ Step2: determine eigenvalue and eigenvector of C
E：principle component base
λ：degree of variance in data
• Step 3: principle component
-determine 1st q component to analysis depends on how
much ability of these components can express variance.
-the higher eigenvalue the more variance it was explained.
• Step4: name and explain each component

PCA model 2
Convert into
(raw data)
(new orthogonal
principal component)
Zi=principle component score(PC score)
-Composed of the coefficient which measure the
contribution of the principle component to each individual
Original waveform data.
-analyzed for group difference using Student’s t-tests.
PCA 1 V.S. PCA 2

Model 1:
Capture information of each parameter independent of time
and show some principle components that can represent the
main profile of these input
useful for observing power coordination

Model 2:
Differentiate difference in each frame of input data
and show these differences by degrees with
principal components.
Sensitive in time domain
Mechanical energy model
Energy analysis of lower extremity

2 methods to determine segment power
◦ Forward dynamic model-based method
◦ Inverse dynamics


Segmental power analysis could not track directly the
work done by calf muscle on the trunk during push off.
Instead, compared ankle joint power to energy passively
transferred through the hip into the trunk via the
proximal linear power term
(Meinders, 2001)

Forward dynamic model combined with segmental
power analysis in lower limb
 quantify the mechanical energy transferred through
the leg by the net joint moments

Neptune et al: examine the effect of individual ankle plantar flexor
muscle contributions to support and forward progression.

Kepple et al: showed pairs of joint moment with opposite energetic
effect work together to balance flow through the segment(ankle
plantar/hip flexor)
eg: hip moment remove energy from leg and simultaneously
ankle moment supplies energy to leg during push-off. Therefore, it
suggested compensation between hip and ankle for muscle
weakness.
Energy flow is a complicate and time-dependent
data.
 Gait data is temporal waveforms including timedependence informaiton.


PCA is an useful tool for capturing timedirection information and give insight into the
dominate performance in power coordination.
Future work of thesis

Use PCA model to find the component of elderly
fallers in gait pattern.
◦ Find motor deficit(eg. Muscle weakness but
peripheral n. intact )elderly fallers to be sujects.
◦ the lower extremity is treated as 3 segament
linked system.(hip/knee/ankle; proximal/distal)
◦ Muscular data would come from saggital
motion(hip flexor/extensor; knee flexor
extensor/ ankle dorsi/plantar flexion)
Thank you for listening!
To be continued~