Balance training improves feedback control of perturbed balance in older adults

TLDR: In this article, the effect of balance training on feedback control after unpredictable perturbations was examined by investigating balance performance, recovery strategy, and muscle synergies, and concluded that balance training improves control of perturbed balance, and reorganizes feedback responses by changing temporal patterns of muscle activation.

Abstract: Recovering balance after perturbations becomes challenging with aging, but an effective balance training could reduce such challenges. In this study, we examined the effect of balance training on feedback control after unpredictable perturbations by investigating balance performance, recovery strategy, and muscle synergies. We assessed the effect of balance training on unipedal perturbed balance in twenty older adults (>65 years) after short-term (one session) and long-term (3-weeks) training. Participants were exposed to random medial and lateral perturbations consisting of 8-degree rotations of a robot-controlled balance platform. We measured full-body 3D kinematics and activation of 9 muscles (8 stance leg muscles, one trunk muscle) during 2.5 s after the onset of perturbation. The perturbation was divided into 3 phases: phase1 from the onset to maximum rotation of the platform, phase 2 from the maximum rotation angle to the 0-degree angle and phase 3 after platform movement. Balance performance improved after long-term training as evidenced by decreased amplitudes of center of mass acceleration and rate of change of body angular momentum. The rate of change of angular momentum did not directly contribute to return of the center of mass within the base of support, but it reoriented the body to an aligned and vertical position. The improved performance coincided with altered activation of synergies depending on the direction and phase of the perturbation. We concluded that balance training improves control of perturbed balance, and reorganizes feedback responses, by changing temporal patterns of muscle activation. These effects were more pronounced after long-term than short-term training.

Figures

Figure 4: Area under the curve of the rate of change of angular momentum after lateral perturbations at three time-points. Top panel a), represents the positive area, and bottom panel b) represents the negative area. Phase 1, 2 and 3 are shown in left, middle and right panel, respectively.

Figure 8: Area under the curve of activation profiles in selected phases after lateral perturbations at three time-points.

Figure 3: Area under center of mass acceleration curve after lateral perturbations at three time-points. Top panel a), represents the positive area, and bottom panel b) represents the negative area. Phase 1, 2 and 3 are shown in left, middle and right panel, respectively.

Figure 7: The muscle weighting and average activation profile shown after medial and lateral perturbations for three phases. Phase1 from perturbation onset to the maximum rotation angle of the platform, phase 2 (return to baseline) from maximum angle to 0-degree rotation angle of the platform and phase3 for 1 s after the platform returned to a 0-degree orientation. Lines reflect Time-points: Pre (— solid line), Post1 (—• dash-dotted line), and Post2 (• dotted line). The red color is assigned to lateral and the black color is assigned to medial perturbations. The baseline values in temporal activation in panel right were subtracted after the factorization to identify suppression and excitation relative to baseline values.

Figure 5: Area under the curve of the center of mass acceleration after medial perturbations at three time-point. Top panel a), represents the positive area, and bottom panel b) represents the negative area. Phase 1, 2 and 3 are shown in left, middle and right panel, respectively.

Figure 6: Area under the curve of the rate of change of angular momentum after medial perturbations at three time-points. Top panel a), represents the positive area, and bottom panel b) represents the negative area. Phase 1, 2 and 3 are shown in left, middle and right panel, respectively.

Full text

Balance training improves feedback
control of perturbed balance in older
adults
Leila Alizadehsaravi
1
, Sjoerd M. Bruijn
1
, Jaap H. van Dieën
1*
1
Department of Human Movement Sciences, Faculty of Behavioural and Movement
Sciences, Vrije Universiteit Amsterdam, Amsterdam, The Netherlands
Corresponding author: prof. dr. Jaap H. van Dieën; j.van.dieen@vu.nl
.CC-BY-NC 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted March 31, 2021. ; https://doi.org/10.1101/2021.03.31.437824doi: bioRxiv preprint

Abstract
Recovering balance after perturbations becomes challenging with aging, but an effective
balance training could reduce such challenges. In this study, we examined the effect of balance
training on feedback control after unpredictable perturbations by investigating balance
performance, recovery strategy, and muscle synergies. We assessed the effect of balance
training on unipedal perturbed balance in twenty older adults (>65 years) after short-term (one
session) and long-term (3-weeks) training. Participants were exposed to random medial and
lateral perturbations consisting of 8-degree rotations of a robot-controlled balance platform.
We measured full-body 3D kinematics and activation of 9 muscles (8 stance leg muscles, one
trunk muscle) during 2.5 s after the onset of perturbation. The perturbation was divided into 3
phases: phase1 from the onset to maximum rotation of the platform, phase 2 from the
maximum rotation angle to the 0-degree angle and phase 3 after platform movement. Balance
performance improved after long-term training as evidenced by decreased amplitudes of center
of mass acceleration and rate of change of body angular momentum. The rate of change of
angular momentum did not directly contribute to return of the center of mass within the base
of support, but it reoriented the body to an aligned and vertical position. The improved
performance coincided with altered activation of synergies depending on the direction and
phase of the perturbation. We concluded that balance training improves control of perturbed
balance, and reorganizes feedback responses, by changing temporal patterns of muscle
activation. These effects were more pronounced after long-term than short-term training.
Keyword: balance training, balance control, feedforward, feedback, counter-rotation, recovery, synergy, aging
.CC-BY-NC 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted March 31, 2021. ; https://doi.org/10.1101/2021.03.31.437824doi: bioRxiv preprint

Introduction
In theory, the nervous system can use two control mechanisms to recover balance after a
perturbation
1
. Reactive or feedback control, occurs after a perturbation and is the only
mechanism available when the nervous system has no prior knowledge of a perturbation
2,3
.
Anticipatory or feedforward control is based on expectations of a perturbation, and aims to
minimize the impact of the perturbation on balance by changing joint orientation or stiffness
prior to a perturbation
4
. Depending on a perturbation's direction and magnitude, feedforward
control is not always sufficient for balance control, and then feedback control comes into play
to regain balance. Effective feedforward control minimizes the effect of perturbations and
reduces the need for feedback control
5,6
.
Three movement strategies are well known to contribute to feedback control of balance after
perturbations: the ankle, counter-rotation, and stepping strategies
7
. The stepping strategy aims
to displace or expand the base of support beyond the projection of the center of mass by
stepping or grabbing a handhold. It is usually seen as a last resort reflecting poorer balance
control, and older adults use it more than younger adults
8–10
. The ankle strategy aims to
accelerate the center of mass towards the base of support through a shift of the center of
pressure, the point of application of the ground reaction force, generated by ankle moments
11
.
The counter-rotation strategy aims to accelerate the center of mass towards the base of support
through horizontal ground reaction forces generated by changes in the angular momentum of
body segments relative to the center of mass
7,11,12
. Thus, these strategies can be differentiated
by distinct kinematics and kinetics but also by distinct patterns of muscle activation reflected in
distinct muscle synergies
13,14
.
In non-stepping balance control, the counter-rotation strategy has been suggested to be
more robust than the ankle strategy
15,16
, and the use of counter-rotation strategies relative to
the ankle increases with age and the magnitude of perturbations
17–20
. Older adults rely on the
counter-rotation strategy at a lower level of challenge than younger adults, even during
unperturbed balancing
17,19
. This presumably helps to secure robust balance control regardless
of age-related sensory errors
21,22
.
Balance training has been shown to result in altered muscle synergies and kinematics after a
perturbation
23,24
. This may reflect improved feedback control but may also reflect improved
feedforward control. Previously, we showed that training of older adults focusing on balance
control on unstable surfaces, improved performance in perturbed and unperturbed balance
tasks
25
. In addition, we found that the duration of co-contraction of muscles around the ankle
increased, and we suggested that this may reflect an improved feedforward control strategy that
.CC-BY-NC 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted March 31, 2021. ; https://doi.org/10.1101/2021.03.31.437824doi: bioRxiv preprint

contributed to performance improvements. Thus, training may have improved feedforward
control resulting in less use of the counter-rotation strategy for balance recovery after a
perturbation. However, in spite of the fact that the training program did not contain sudden,
unpredictable perturbations, the challenging exercises used in training may also have improved
feedback balance control, in which case one might expect the more effective counter-rotation
strategy to be used more after training. In this study, we investigated the effects of training on
kinematics and muscle synergies of balance recovery after perturbations in more detail to
improve our understanding of training effects on feedback control of balance in older adults.
Methods
The data collection and training were described earlier
25
, here we provide a brief summary.
In this study twenty older adults (71.9±4.09 years old) participated. All participants provided
written informed consent prior to participation, and the ethical review board of the Faculty of
Behavioural and Movement Sciences, Vrije Universiteit Amsterdam, approved the
experimental procedures (VCWE-2018-171).
Training consisted of balancing on balance boards and foam pads. The first training session
was completed individually (30 minutes), and subsequently, a 3-weeks training program was
completed in groups of 6-8 participants (45x3 minutes per week). We gradually increased the
challenge of exercises by reducing hand support, moving from bipedal to unipedal stance, using
more unstable support surfaces, and adding perturbations such as catching and throwing a ball
and reducing visual input.
We assessed balance recovery with participants in unipedal stance on their dominant leg on
a robot-controlled platform (HapticMaster, Motek, Amsterdam, the Netherlands). Participants
performed 5 trials of a perturbed unipedal balance task, in which 12 random perturbations (6
medial and 6 lateral) were induced during 50-60 seconds. The platform rotated over a sagittal
axis in the medial or lateral direction (amplitude of 8°) in random order. Participants were
given two minutes rest between trials and a randomized 3-5 seconds rest period between
perturbations within the trial. Participants were asked to fix their vision on a target in front of
them. Full-body 3D kinematics were tracked by one Optotrak camera array (Northern Digital,
Waterloo, Canada). Surface electromyography (EMG) data were recorded from nine unilateral
muscles of the dominant leg: tibialis anterior (TA), vastus lateralis (VL), lateral gastrocnemius
(GsL), soleus (SOL), peroneus longus (PL), rectus femoris (RF), biceps femoris (BF) and gluteus
medius (GlM) and erector spinae (ES) muscles (TMSi, Twente, The Netherlands). We collected
the data at baseline (Pre), after one training session (Post1), and after ten training sessions
(Post2).
.CC-BY-NC 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted March 31, 2021. ; https://doi.org/10.1101/2021.03.31.437824doi: bioRxiv preprint

Data analysis
Sixty perturbations per participant per time-point (30 medial and 30 lateral) were used to
calculate all variables.
Perturbation Onset
The onset of the perturbations was detected through the platform's rotation angle after
synchronizing the platform, kinematics, and EMG data. Medial perturbations were defined
when the platform started to rotate such that the big toe moved downward (eversion) and lateral
when the big toe moved upward (inversion), and this was consistent for right- and left-leg
dominant participants. A time window from 0.5 s before the onset of the perturbation to 2.5 s
after the onset was selected for further analysis of all variables. For all variables 0.5 s baseline
(from the start of the window until perturbation onset) was subtracted. Kinematics data were
ensemble-averaged first over perturbations within a trial and then over trials per participant.
The selected window was divided into three sub-windows; phase1 from perturbation onset to
the maximum rotation angle of the platform, phase 2 (return to baseline) from maximum angle
to 0-degree rotation angle of the platform and phase3 for 1 s after the platform returned to a
0-degree orientation.
Balance recovery, performance and strategy
We averaged the time series of center of mass displacement (CoM [m]), velocity (vCoM
[m/s]) and acceleration (aCoM [m/s
2
]) in the frontal plane over all trials at a given time-point
per subject. We calculated the positive and negative areas under the center of mass acceleration
curve as an indicator of balance performance
26
. Next, we calculated total body angular
momentum [kg.m
2
/s], its integral after division by the instantaneous moment of inertia to
obtain a description of body orientation [degree], and the rate of change in total body angular
momentum (time derivative of the total body angular momentum [kg.m
2
/s
2
]). We calculated
the positive and negative areas under the curve of the rate of change of angular momentum as
a second indicator of performance. The positive and negative areas were estimated separately
for the three phases per direction of perturbation. The counter-rotation strategy is used when
the rate of change in angular momentum accelerates the center of mass towards the base of
support. Independent of this, angular momentum may be changed to regain upright body
orientation. To assess how angular momentum changes were used, we compared the direction
and timing of changes in the rate of change of angular momentum with CoM position and with
body orientation.
.CC-BY-NC 4.0 International licensemade available under a
(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is
The copyright holder for this preprintthis version posted March 31, 2021. ; https://doi.org/10.1101/2021.03.31.437824doi: bioRxiv preprint

Frequently asked questions

Q1. What have the authors contributed in "Balance training improves feedback control of perturbed balance in older adults" ?

In this study, the authors examined the effect of balance training on feedback control after unpredictable perturbations by investigating balance performance, recovery strategy, and muscle synergies. The authors concluded that balance training improves control of perturbed balance, and reorganizes feedback responses, by changing temporal patterns of muscle activation. CC-BY-NC 4. 0 International license made available under a ( which was not certified by peer review ) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. 

Q2. What is the preferred orientation of the CoM?

While the CoM could be maintained over the base of support with opposite orientations of the upper and lower body oriented, a vertical orientation of both segments seems to be preferred and would of course be less demanding. 

Q3. How was the onset of the perturbations detected?

The onset of the perturbations was detected through the platform's rotation angle after synchronizing the platform, kinematics, and EMG data. 

Q4. What is the effect of training on the negative area under the acceleration curve?

The negative area under the acceleration curve in phase 1, in the direction of the platform rotation, was affected by training (F2,38 = 3.53, p = 0.039). 

Q5. What was the effect of training on the positive area under the rate of change of angular?

In phase 2, although the activation profile was mainly above baseline (excitation), there was a significant effect of training on the negative area under the curve (inhibition) of synergy 1 (F2,38 = 3.62, p = 0.036). 

Q6. What is the effect of training on the positive area under the acceleration curve?

In phase 1, the initial negative area under the rate of change of angular momentum curve, in the direction of the platform rotation, was affected by training (F2,38 = 4.52, p = 0.017). 

Q7. What is the effect of the reorientation of the body after training?

This reorientation of the body was better tuned after training, i.e., the corrective change in angular momentum had a smaller area under the rate of change of angular momentum curve, resulting in less overshoot. 

Q8. What was the effect of training on the negative area under the curve of synergy 4?

Training also affected the negative area under the curve (inhibition) of synergy 4 in phase 2 (F2,38 = 4.31, p = 0.02), although the average activation in this phase was positive. 

Q9. What was the effect of the training on the positive area under the rate of change of ang?

In phase 1, the positive area under the acceleration curve, in the direction of the platform rotation, was affected by training (F2,38 = 8.61, p < 0.001). 

Q10. What is the effect of training on balance?

The significant improvement in balance performance of phase 2 after perturbations indicates that the feedback control of balance improved more notably after long-term training. 

Q11. What was the effect of training on the positive area under the curve of synergy 3?

In phase 3, there was an effect of training on the positive area under the curve (excitation) of synergy 3 (F2,38 = 3.67, p = 0.035). 

Q12. What was the effect of training on the negative area under the rate of change of angular?

In phase 2, the negative area under the rate of change of angular momentum curve, in the direction of the platform rotation back to horizontal, was also affected by training (F2,38 = 7.26, p = 0.002). 

Q13. What is the effect of training on the area under the rate of change of angular momentum?

Post-hoc testing showed that area under the acceleration curve did not change after short-term (p = 0.07) but decreased afterLateral perturbation Rate of change of angualar momentumlong-term training (t = 4.14, p < 0.001; Figure 5.a, left panel). 

Q14. What did the authors find out about the effects of training on the balance of the ankle?

Previously the authors reported that training decreased mean absolute center of mass velocity and increased ankle muscle co-contraction in perturbedMedial perturbationunipedal balancing25.