Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Nov 1.
Published in final edited form as: Exp Brain Res. 2021 Sep 3;239(11):3327–3341. doi: 10.1007/s00221-021-06202-9

Enhanced error facilitates motor learning in weight shift and increases use of the paretic leg during walking at chronic stage after stroke

Seoung Hoon Park a,b, Chao-Jung Hsu a, Weena Dee a, Elliot J Roth a,b, William Z Rymer a,b, Ming Wu a,b,c
PMCID: PMC8541925  NIHMSID: NIHMS1744091  PMID: 34477919

Abstract

The purpose of this study was to determine whether the application of lateral pelvis pulling force toward the non-paretic side during the stance phase of the paretic leg would enhance forced use of the paretic leg and increase weight shift toward the paretic side in stroke survivors. Eleven chronic stroke survivors participated in two experimental sessions, which consisted of 1) treadmill walking with the application of “pelvis resistance” or “pelvis assistance” and 2) overground walking. During the treadmill walking, the laterally pulling force was applied during the stance phase of the paretic leg toward the non-paretic side for the “pelvis resistance” condition or toward the paretic side for the “pelvis assistance” condition during the stance phase of the paretic leg. After force release, the “pelvis resistance” condition exhibited greater enhancement in muscle activation of hip ABD, ADD, and SOL and greater improvement in lateral weight shift toward the paretic side, compared with the effect of the “pelvis assistance” condition (P<0.03). This improved lateral weight shift was associated with the enhanced muscle activation of hip ABD and ADD (R2=0.67, P=0.01). The pelvis resistance condition also improved overground walking speed and stance phase symmetry when measured 10-min after the treadmill walking (P=0.004). In conclusion, applying pelvis resistance forces to increase error signals may facilitate motor learning of weight shift toward the paretic side and enhance use of the paretic leg in chronic stroke survivors. Results from this study may be utilized to develop an intervention approach to improve walking in stroke survivors.

Keywords: Stroke, Locomotion, Forced use, Weight shift, Pelvis resistance, Constraint-induced movement therapy

1. INTRODUCTION

Many individuals with hemiparetic stroke across subacute and chronic stage show impairment in walking, characterized by asymmetrical gait patterns and slow walking speed (O’lney and Richards 1996; Roerdink et al. 2009). Since walking ability is a deciding factor in ability to perform independent activities of daily living and high quality of life (Bohannon et al. 1988; Almkvist Muren et al. 2008; Gordon et al. 2013), improving walking ability has been set as a crucial goal of stroke rehabilitation. Treadmill training is one approach that has been utilized in clinical settings to improve walking function in individuals with chronic and subacute stroke (Barbeau 2003; Plummer et al. 2007). Recent systematic reviews indicated that treadmill training may induce significant improvements in walking speed and endurance in individuals with chronic and subacute stroke (Mehrholz et al. 2017; Nascimento et al. 2021). While these improvements in chronic stroke survivors were statistically significant (Nindorera et al. 2021), the functional gains are partially due to increased compensatory strategies from the non-paretic leg (Ardestani et al. 2019). A possible reason for this is that current conventional locomotor treadmill training does not focus on the weakness of the paretic leg but rather allows stroke survivors to walk with compensatory strategies, such as relying more on the non-paretic leg (Raja et al. 2012; Roelker et al. 2019). For instance, many chronic stroke survivors show insufficient weight shift toward the paretic side during treadmill walking (Hsu et al. 2017; Wu et al. 2017). This phenomenon may hinder the initiation of the swing of the non-paretic leg. Previous study suggested that impairment in weight shift toward the paretic side is significantly correlated with slow walking speed in post-acute stroke survivors (Kamono and Ogihara 2018). Repeated practice in this gait pattern may simply reinforce this compensatory strategy that could lead to less effective in improving motor function of the paretic leg. As a consequence, the effect of conventional treadmill training on walking function in individuals post-stroke could be suboptimal. Therefore, there is a clear need to develop an intervention strategy that can force or promote stroke survivors to use the paretic leg more and attenuate compensatory strategy from the non-paretic side during treadmill training.

Constraint induced movement therapy (CIMT) may reverse compensatory walking strategies by promoting forced used of the paretic leg (Wolf et al. 2008; Hsu et al. 2017). The CIMT has been commonly used to induce forced use of the paretic upper limb in individuals post-stroke (Taub et al. 2006; Wolf et al. 2008). CIMT protocol consists of massed task specific practice with the affected limb; training with a behavior technique termed shaping; transfer package; and constraint of the unaffected limb (Taub et al. 2013b). Results showed that forced use of the paretic limb may reverse the “learned nonuse” of the paretic limb (Taub et al. 1994). It has been suggested that there are two potential mechanisms that underlie the treatment effects observed with CIMT: overcoming learned nonuse and use-dependent plastic brain reorganization (Taub et al. 2013a). The promising results of CIMT on motor recovery of the paretic arm also inspired some researchers to transfer this technique to the lower limb interventions, but it has been challenging to apply this training paradigm to locomotor training for stroke survivors since both legs must be used during locomotion-related tasks. Results from our previous studies suggest that applying corrective pulling force to the pelvis toward the paretic side during walking (i.e., pelvis assistance force) may induce enhanced muscle activity of the paretic leg and improve gait symmetry of individuals post chronic stroke (Hsu et al. 2017). Repeated application of the pelvis assistance force toward the paretic side might also induce motor adaptation, i.e., participants showed improved weight shift toward the paretic side during the early adaptation period, but the weight shift toward the paretic side gradually returned to a level that was comparable to baseline during the mid-late adaptation period. This observation is consistent with a previous study in individuals post chronic stroke during split-belt treadmill walking, which showed that the participants with stroke adapted toward their baseline asymmetry (Malone and Bastian 2014). Participants showed an aftereffect with reduced weight shift toward the paretic side after release of the pelvis assistance force (Wu et al. 2017).

On the other hand, results from a novel earlier study suggest that enhancing error may accelerate motor learning in reaching of the paretic arm in chronic stroke survivors (Patton et al. 2006). In line with this, results from lower limb studies indicate that increasing error may enhance motor learning for improving gait symmetry in chronic stroke survivors (Reisman et al. 2007; Savin et al. 2014; Yen et al. 2015) or enhance step length in human with spinal cord injury (Houldin et al. 2011; Yen et al. 2013). It remains unclear as to whether applying pelvis resistance force to increase error signals instead of assistance force during walking would facilitate motor learning in weight shift toward the paretic side in stroke survivors. There is a possibility that applying a pulling force toward the “non-paretic” side during the stance phase of the paretic leg (i.e., pelvis resistance force) may be more effective in enhancing lateral weight shift toward the paretic side during walking in stroke survivors. Specifically, walking practice with pelvic resistance force toward the non-paretic side may force hemiparetic stroke survivors to overcome the resistive force in order to shift their body weight toward the “paretic” side to initiate a non-paretic leg swing (Grieve and Gear 1966). Repeating these actions against pelvic resistance force during standing phase of the paretic leg may further enhance the use of the paretic leg and potentially improve lateral weight shift toward the paretic side potentially through use-dependent motor learning mechanisms (Diedrichsen et al. 2010; Wu et al. 2019).

The purpose of this study, therefore, was to determine whether applying a pulling force to the pelvis toward the non-paretic side (i.e., pelvis resistance) versus a pulling force toward the paretic side (i.e., pelvis assistance) during the stance phase of the paretic side can induce improved lateral weight shift toward the paretic side and enhance muscle activity of the paretic leg in chronic stroke survivors. Additionally, we wanted to determine whether the improvement in weight shift toward the paretic side and enhanced use of the paretic leg would be transferred later to overground walking. We hypothesized that 1) treadmill walking with pelvis resistance force would result in enhanced muscle activation of the paretic leg during stance and in improved lateral weight shift toward the paretic side; and 2) the improved motor control of the paretic leg gained from treadmill practice would result in an increase in overground walking speed and improved gait symmetry, compared with the effects of the pelvis assistance force.

2. METHODS

2.1. Participants

Eleven individuals (60.0 ± 4.4 years, 6 females) with chronic stroke (>6 months) were recruited to participate in this study from the outpatient clinics of the Shirley Ryan AbilityLab or from its stroke patient registry (see Table 1). The inclusion criteria were: 1) 21-75 years old; 2) a single unilateral, supratentorial stroke of either ischemic or hemorrhagic etiology, confirmed by radiographic imaging; 3) showing weakness/paresis of the affected leg; and 4) ability to stand and walk independently (>10 m) without physical assistance from a physical therapist, and with the use of assistive devices or ankle-foot orthoses if needed. Exclusion criteria were: 1) brainstem or cerebellar stroke; 2) a score of <24 on the Mini Mental State Examination (Folstein et al. 1983); 3) other neurological conditions, cardiorespiratory/metabolic disorders, or orthopedic conditions affecting ambulatory ability; 4) uncontrolled hypertension (systolic >200 mg Hg, diastolic >110 mm Hg); 5) botulinum toxin injection within the prior 6 months; and 6) inability to tolerate 30 min of treadmill walking with body weight supported as needed. The Northwestern University Medical School Institutional Review Board approved all procedures and all participants signed informed consent before data collection.

Table 1.

Demographic information for the participants.

P Sex Age (y) Weight (kg) Height (m) Post-injury (y) Paretic side Brace Assistive device Self-selected comfortable speed (m/s) Assist/resist force
N %BW
1 F 57 94.3 1.65 8 R None None 0.48 75 8
2 M 67 78.5 1.80 4 R AFO SPC 0.46 46 6
3 M 58 75.7 1.78 10 L AFO None 0.84 74 10
4 M 55 81.6 1.75 9 R AFO None 0.64 80 10
5 F 63 77.1 1.55 27 L AFO Rollator 0.32 68 9
6 F 62 56.7 1.68 2 R AFO SPC 0.38 50 9
7 M 67 71.2 1.73 5 L AFO none 0.5 56 8
8 F 65 69.9 1.63 7 L AFO SPC 0.73 48 7
9 M 63 102.1 1.85 17 R None SPC 0.44 60 6
10 F 60 94.3 1.65 1 R None None 0.52 65 7
11 F 57 61.2 1.68 7 R AFO SPC 0.55 42 7
12 M 53 88.0 1.70 20 R AFO None 0.65 69 8
Open in a new tab

Abbreviations: P, participant; %BW, percentage of body weight; M, male; F, female; R, right; L, left; AFO, ankle foot orthosis; SPC, single point can.

2.2. Apparatus

We used a custom designed cable-driven robotic system, which was mounted over a treadmill, to apply pelvic resistance force toward the non-paretic side or to apply pelvic assistance force toward the paretic side during walking (Figure 1A & B). Specifically, the robotic system was composed of 2 nylon-coated stainless-steel cables (3mm, in diameter) driven by 2 motors (AKM 33H, Kollmorgen, Radford, VA) through cable spools located at both sides of the treadmill (Wu et al. 2011). In this study, we used a motor and a cable at one side of treadmill to apply controlled pulling force to the pelvis, which was either resistant or assistant to the pelvis movements in the mediolateral direction, depending on which side was affected and which condition was tested.

Figure 1.

Figure 1.

Open in a new tab

Experimental setup and protocol. A) Participants walked on a treadmill with laterally pulling force either toward the non-paretic side (red arrow; pelvis resistance condition) or toward the paretic side (blue arrow; pelvis assistance condition), applied by a cable-driven robotic system. While walking on the treadmill, muscle activation from the paretic leg and position signals were recorded. B) The timing of applying laterally pulling force toward the non-paretic side (NP; red arrow) or the paretic side (P; blue arrow) was the early-to-mid stance phase of the paretic leg. C) All participants visited the lab once to complete the two testing sessions (i.e., pelvis resistance vs. assistance condition), which were randomly ordered (n = 6, resistance first, → then, assistance; n = 5, assistance first → then, resistance). A 10-min seated break was inserted between the two testing sessions. D) During each session, participants performed the treadmill walking task: 1-min baseline treadmill walking; 7-min treadmill walking with either lateral resistance or assistance force (adaptation); 1-min treadmill walking (post-adaptation); 2-min standing break; 5-min treadmill walking with either lateral resistance or assistance force (re-adaptation). Also, participants conducted the overground walking task before treadmill walking, immediately after treadmill walking, and 10 min after treadmill walking. Abbreviation: B, baseline; A1-5, adaptation periods; PA1-2, post-adaptation periods.

Specifically, for the assistance condition, we attached one cable to the participant’s pelvis on the paretic side through a waist belt and used a pulley system to deliver lateral assistance force to the pelvis toward the paretic side during the stance phase of the paretic leg (i.e., the direction of lateral force was congruent with the movement of the pelvis in the mediolateral direction to facilitate weight shift toward the paretic side). For the resistance condition, a cable was attached to the pelvis on the non-paretic side and was used to deliver a lateral pulling force to the pelvis toward the non-paretic side during the stance of the paretic leg (i.e., the direction of the lateral pelvis force applied was opposite to the movement of the pelvis in the mediolateral direction). Further, a set of customized 3-dimensional position sensors was attached to participants’ ankles and pelvises to collect position signals of the pelvis and ankle (Figure 1A) (Yen et al. 2012). The sensor consists of a detector rod and three potentiometers. Specifically, two potentiometers (model: P2201, Novotechnik, Southborough, MA, USA) were used to measure rotational movements of the rod in the anteroposterior and mediolateral directions. The other linear potentiometer (model: SP-2, Celesco, Chatsworth, CA, USA) was used to measure the linear displacement of the rod in the vertical direction. One end of the sensor was affixed to a fixed aluminum frame that was located at the side of treadmill, and the other side of sensor was attached to the leg through a strap or to the pelvis using a belt. The robotic system was controlled by a custom-written program in LabVIEW (National Instrument, Austin, TX, USA) on a personal computer. The controller used ankle position signals to trigger assistance/resistance force at a targeted phase of gait, i.e., early-to-mid stance phase of the paretic leg (Figure 1B).

2.3. Experimental protocol

All participants were tested under two conditions (i.e., pelvis resistance and pelvis assistance) using an across-over design within one lab visit. Each condition was conducted in a separate experimental session, and the order of the two sessions was randomized across participants (Figure 1C; six participants were tested in resistance condition first, then, assistance condition; five participants were tested in assistance condition first, then, resistance condition). There was a 10-minute seated break between the two testing sessions. A lateral pulling force was applied to the pelvis either toward the non-paretic for the resistance condition or toward the paretic side for the assistance condition during the time when the paretic leg was in contact with the treadmill belt (i.e., starting from ~100ms following the initial contact of the paretic leg for 400ms). This early-to-mid stance phase was selected because during this period of time the non-paretic leg was mostly in the swing phase and therefore the pulling force could force participants to use more of their paretic leg and mitigate the compensatory movement from the non-paretic leg. The magnitude of force was set at 8% of body weight, which was determined based on previous studies (Hsu et al. 2017; Park et al. 2020) and was adjusted depending on each participant’s tolerance (Table 1). Specifically, we gradually increased the magnitude of force from 0 to 8% of body weight (%BW) while participants walked on a treadmill at their comfortable speed. An experienced physical therapist who closely watched the walking performance of participants determined the level of the force based on the feedbacks from the participants and her clinical judgement. Each session included the following procedures (Figure 1D): 1) baseline overground walking; 2) treadmill walking without the application of pelvis resistance or assistance force for 1 min (baseline); 3) treadmill walking with the application of pelvis resistance or assistance force for 7 min (adaptation); 4) treadmill walking without the force for additional 1 min (post-adaptation); 5) standing break for 2 min; 6) treadmill walking with the application of pelvis resistance or assistance force for additional 5 min (re-adaptation), which was used in order to re-induce motor adaptation; 7) overground walking immediately post treadmill walking; 8) 10-min seated break; and 9) overground walking. Participants walked over ground on the GAITRite® mat (CIR Systems, Inc., Franklin, NJ, USA) for three trials in each condition. We transported participants using a wheelchair from the treadmill to the GAITRite® mat to reduce the potential washout of motor skills during the transition period. The treadmill walking speed was set at a self-selected comfortable speed for each participant, which was determined at the beginning of the testing session. This selection remained the same for all testing sessions. An overhead harness was used for the purpose of safety only (no body weight support was provided). Participants were allowed to hold onto the front handrail using their hands during treadmill walking for safety. Before data collection, participants walked on the treadmill at their maximum speed for 30 strides during which EMG activity was recorded and was used to normalize the integrated EMGs. For these participants who wear AFO during their regular walking, they were allowed to use AFO during all experimental sessions. For these participants who need assistive device, such as cane, during regular walking, they were allowed to use it during overground walking test, but were requested to be consistent across testing sessions.

2.4. Data collection

We recorded muscle activity of tibialis anterior (TA), medial gastrocnemius (MG), soleus (SOL), rectus femoris (RF), vastus medialis (VM), medial hamstring (MH), hip abductor (ABD; gluteus medius), and hip adductor (ADD; adductor magnus) in the paretic leg using surface EMG electrodes (BagnoliTM, Delsys, Boston, MA, USA). EMG signals were amplified (×1000) and band-pass filtered (20-450 Hz) using Bagnoli-16 Amplifier (Delsys, MA, USA), and then sampled with an A/D board (National Instruments, Austin, TX, USA) at 500 Hz using a custom-written program in LabVIEW (National Instruments, Austin, TX, USA). Ankle and pelvis positions were sampled at 500 Hz with customized 3-dimensional position sensors (Figure 1A) using the customized program in LabVIEW. Gait speed and other spatiotemporal gait parameters during overground walking were calculated using the software from the GAITRite®.

2.5. Data analysis

All EMG and kinematic data were analyzed using custom-written programs in MATLAB (MathWorks, MA, USA). EMG data were high-pass filtered at 10 Hz, notch filtered from 59-61 and 119-121 Hz (i.e., to remove electrical current noise, 60 Hz and its harmonic frequency, 2*60=120 Hz), rectified, and smoothed using a low-pass filter at 20 Hz (fourth-order Butterworth). The EMG data were segmented into different gait cycles based on ankle position data and were normalized to peak values of each muscle activity while participants walked at their maximal walking speed, and then were multiplied by 100 (unit%). Then, the integrals of muscle activities during stance phase of gait were calculated. Ankle and pelvis position signals were low-pass filtered at 10 Hz. To calculate pelvis mediolateral displacement of each gait cycle, we subtracted the mid-point between two ankle positions at the initial contact of each leg from the mid-point between paretic and non-paretic pelvis positions. The mid-point of two ankle positions at the initial contact was defined as a reference point for each gait cycle because of step-to-step variation of foot position in the mediolateral direction during treadmill walking. Then, we quantified the peak of pelvis lateral displacement during stance phase of the paretic leg as weight shift toward the paretic side (Hsu et al. 2019; Park et al. 2020). In addition, we quantified the number of paretic leg steps with weight shift toward the paretic side exceeding the baseline level (i.e., for the steps when the peak of pelvis displacement was greater than 1 standard deviation + mean of last 30 steps during baseline) among the last 100 steps of the adaptation period.

EMG integrals and weight shift during the following 8 subintervals were calculated and compared across two testing conditions or different time periods within each condition (Park et al. 2020). The subintervals were: the last 30 steps during the baseline (B); the first 5 steps (A1), 5 steps of the first quarter (A2), the middle 5 steps (A3), 5 steps of the third quarter (A4), and the last 5 steps (A5) during the adaptation period; the first (PA1) and last 5 steps (PA2) during the post-adaptation period. To compare changes in integrated EMG and weight shift between two testing conditions, the average of each variable during baseline (i.e., B) was subtracted from that during late post-adaptation period (i.e., PA2).

The symmetry of stance time during overground walking was quantified as following (Lauziere et al. 2014):

Symmetry index=(PareticNon-paretic)×100 (1)

2.6. Statistical analysis

A two-way ANOVA with repeated measures was used to test interactions between condition (i.e., 2 conditions: resistance vs. assistance) and time [3 time points: baseline, PA1, & PA2 for treadmill walking; baseline (i.e., pre-treadmill walking), immediately post treadmill walking (P), and 10-minute post treadmill walking (10 min P) for overground walking] on the integrated EMG, weight shift, and overground walking speed.

A one-way ANOVA with repeated measures was used to compare the integrated EMG (Figure 4, the middle column of graphs in each box), weight shift (Figure 5B), stance time symmetry (Figure 7A) during overground walking, and overground walking speed at different time periods (3 time points: baseline, PA1, & PA2 for treadmill walking; baseline (i.e., pre-treadmill walking), P, & 10 min P for overground walking) within each experimental condition. If a significant main effect was identified, post-hoc tests with the Bonferroni adjustment were further conducted to identify significant differences.

Figure 4.

Figure 4.

Open in a new tab

Muscle activation of the paretic leg during the treadmill walking for the pelvis resistance and assistance conditions. Data shown in the left column of each graph box were the average of integrated EMG during baseline (B), the adaptation period (i.e., A1-A5 from the early adaptation to the late adaptation periods), and the post-adaptation period (PA1 and PA2) across participants. Data shown in the middle column of each graph box were the average of integrated EMG during baseline (B), and the post-adaptation period (PA1 and PA2) across participants. Data shown in the right column of each graph box were the average of the change in integrated EMG from baseline to the post-adaptation period (PA1 and PA2). Abbreviation: B, baseline; A1-5, adaptation periods; PA1, early post-adaptation period; PA2, late post-adaptation period. Asterisks (*) indicate significant difference.

Figure 5.

Figure 5.

Open in a new tab

Lateral weight shift toward the paretic side during treadmill walking. Data shown in column A were average of lateral weight shift during baseline (B), the adaptation period (A1 to A5, from the early adaptation period to the late adaptation period), and the post-adaptation period (PA1 and PA2). Data shown in column B were average of lateral weight shift during baseline (B) and the post-adaptation period (PA1 and PA2). Data shown in column C were the average of change in weight shift from baseline to the post-adaptation period (PA1 and PA2). The correlation between the change in weight shift toward the paretic side from baseline to the late post-adaptation period and predicted change in weight shift based on muscle activity of ABD and ADD is shown in column D. Abbreviation: B, baseline; A1-5, adaptation periods; P1, early post-adaptation period P2, late post-adaptation period. Asterisks (*) indicate significant difference.

Figure 7.

Figure 7.

Open in a new tab

Stance time symmetry during overground walking. A. Average of stance time symmetry during baseline, immediately post treadmill walking (P), and 10-minute post treadmill walking (10 min P) for both the resistance and assistance conditions. B. Average of the change in stance time symmetry (from baseline to 10-minute post treadmill walking). C. Average of the change in stance time of the paretic leg and non-paretic leg from baseline to 10-minute post treadmill walking. Asterisks (*) indicate significant difference.

Paired t tests were used to compare 1) integrated EMG (Figure 4, the first column of graphs in each box) and weight shift (Figure 5A) between the two experimental conditions for each time of periods separately, 2) changes in integrated EMG (Figure 4, the third column of graphs in each box), weight shift (Figure 5C), stance time and its symmetry (Figure 7B&C) during overground walking, and overground walking speed, and 3) the number of paretic leg steps with weight shifts exceeding the baseline mean among the last 100 steps of the adaptation period (Figure 6A) between the two testing conditions.

Figure 6.

Figure 6.

Open in a new tab

A. Number of paretic leg steps with lateral weight shifts exceeding the baseline level for the resistance and assistance conditions. B. The correlation between the change in weight shift from baseline to late post-adaptation period (PA2) and the number of steps when the weight shift toward the paretic side exceeding baseline during the late adaptation period (the last 100 steps of the adaptation period). Asterisks (*) indicate significant difference.

A backward multiple linear regression model (Figure 5D) was used to establish statistical models to predict the improvement in lateral weight shift toward the paretic side (i.e., change from the baseline to the late post-adaptation period) from the integrated EMG of all muscles. A linear regression model (Figure 6B) was used to determine the relation between the improvement in weight shift toward the paretic side from baseline to late post-adaptation period and the number of paretic leg steps with weight shift exceeding the baseline among the last 100 steps of the adaptation period.

For the treadmill walking, we selected the time point PA1 (early post-adaptation) to examine the immediate aftereffect of the motor adaptation induced by the treatment (i.e., resistance or assistance) on the integrated EMG and weight shift, and the time point PA2 (late post-adaptation) to examine the retention of that aftereffect. Specifically, we focused on the retention of motor adaptation during the late post-adaptation period because it has more potential clinical significance.

The IBM SPSS Statistics 20.0 statistical package (IBM Corp., Armonk, NY, USA) was used for analyses. The alpha level for all statistical tests was 0.05. Data are reported as mean ± standard deviation in the text and mean ± standard error of the mean in the figures.

3. RESULTS

3.1. Weight shift toward the paretic side during treadmill walking

Lateral weight shift toward the paretic side during treadmill walking from one typical participant for the resistance and assistance conditions is shown in Figure 3. For the resistance condition, the application of lateral pulling force toward the non-paretic side induced an immediate decrease in weight shift toward the paretic side (i.e., the participant showed less weight shift toward the paretic side) during the early adaptation period. However, over several steps, it returned to a level that was comparable to baseline, and then after ~220 steps, it increased to a level that was greater than baseline for the remainder of adaptation period. Following the resistance force release, the participant showed an aftereffect consisting of increased weight shift toward the paretic side during the early post-adaptation period. Then, the aftereffect washed out within ~12 steps. Afterwards, the weight shift toward the paretic side gradually improved again during the late post-adaptation period. For the assistance condition, the application of lateral assistance force toward the paretic side led to an immediate improvement in lateral weight shift toward the paretic side during the early adaptation period. It gradually returned to a level that was comparable to baseline, and further decreased during the period of ~80-150 steps. Then, it remained at a level that was less than baseline for the remainder of the adaptation period. Following the assistance force release, the participant showed an aftereffect that consisted of a decreased weight shift toward the paretic side during the early post-adaptation period, and then it gradually returned to a level that was comparable to baseline within ~14 steps during the late post-adaptation period.

Figure 3.

Figure 3.

Open in a new tab

Stride-by-stride lateral weight shift toward the paretic side from one representative participant for the pelvis resistance and assistance conditions.

The group average of weight shift toward the paretic side (i.e., peak of pelvis lateral displacement during stance phase of the paretic leg) is shown in Figure 5. Participants showed greater weight shift toward the paretic side during the late adaptation period, and early and late post-adaptation periods for the pelvis resistance condition than for the pelvis assistance condition (i.e., A4: t = −2.48, P = 0.03; PA1: t = −4.62, P = 0.001; PA2: t = −7.58, P < 0.001). They also demonstrated comparable weight shift toward the paretic side during baseline and the early adaptation period (i.e., baseline, A1-3, & A5; P > 0.05).

A two-way ANOVA with repeated measures revealed that the condition × time interaction was significant for weight shift toward the paretic side during treadmill walking (F = 16.69, P < 0.001). There was significant main effect of condition for weight shift toward the paretic side (F = 29.84, P < 0.001) whereas there was no significant main effect of time for weight shift (F = 3.12, P = 0.07).

A one-way ANOVA with repeated measures indicated that the application of pelvis resistance had significant impact on weight shift toward the paretic side during treadmill walking (F = 13.85, P < 0.001). Post-hoc analysis using the Bonferroni correction (3 comparisons conducted) showed that participants showed significant increase in weight shift toward the paretic side from baseline to the early post-adaptation period (P = 0.01), and to the late post-adaptation period (P < 0.001) for the pelvis resistance condition. Similarly, the application of the pelvis assistance had a significant impact on weight shift toward the paretic side (F = 3.94, P = 0.04). However, post-hoc analysis showed that participants showed significant decreases in weight shift toward the paretic side from baseline to the early post-adaptation period (P = 0.02) for the pelvis assistance condition (Figure 5B).

Changes in weight shift from baseline to the early post-adaptation period (t = −4.75, P = 0.001) and to the late post-adaptation period (t = −5.90, P = 0.0002) were significantly different between the two conditions and the direction of those changes was opposite across the two conditions. Specifically, the increase in weight shift from baseline to the late post-adaptation period was significantly related to the increase in muscle activation of ABD and ADD (R2 = 0.67, DW = 1.81, P = 0.01; Figure 5D).

The number of paretic leg steps with weight shift toward the paretic side exceeding the baseline level during the late adaptation period (i.e., the last 100 steps before load release) was significantly greater for the resistance condition than for the assistance condition (t = −3.94, P = 0.003). In the resistance condition, participants who had greater number of steps with weight shift toward the paretic side exceeding the baseline level also exhibited greater improvement in weight shift toward the paretic side from baseline to the late post-adaptation period (R2 = 0.38, DW = 2.75, P = 0.042).

3.2. Neuromuscular activation of the paretic-leg muscles during treadmill walking

Muscle activation of the paretic leg from one typical participant for the resistance and assistance conditions is shown in Figure 2. For the resistance condition, this participant showed an increase in muscle activity of ABD, MH, and SOL during the early adaptation period and the late post-adaptation period, in comparison to baseline. For the assistance condition, this participant showed an increase in ABD muscle activation during the early adaptation period, and showed modest changes in muscle activity of ABD, MH and SOL during the late post-adaptation period, in comparison to baseline.

Figure 2.

Figure 2.

Open in a new tab

Representative muscle activation of paretic ABD, MH, and SOL during gait cycles of the paretic leg from one representative participant.

The group average of integrated EMG from the muscles in the paretic leg is shown in Figure 4. A two-way ANOVA with repeated measures revealed that the condition × time interaction was significant for muscle activation of SOL (F = 9.09, P = 0.002), MH (F = 4.40, P = 0.03), and ABD (F = 5.29, P = 0.01) but was not significant for the activation of other muscles (TA: F = 0.14, P = 0.87; MG: F = 2.92, P = 0.08; VM: F = 0.38, P = 0.69; RF: F = 0.88, P = 0.43; ADD: F = 0.25, P = 0.78). There was no significant main effect of condition for the activation of all muscles (F < 2.01, P > 0.18). There was significant main effect of time for MG (F = 6.62, P = 0.01), SOL (F = 7.17, P = 0.004), and ABD (F = 4.05, P = 0.03) muscle activation but was not significant for the activation of other muscles (F < 2.68, P > 0.09).

For the pelvis resistance condition, a one-way ANOVA with repeated measures revealed that muscle activation of MG (F = 5.91, P = 0.01), SOL (F = 12.54, P = 0.0003), MH (F = 5.29, P = 0.01), and ABD (F = 7.19, P = 0.004) differed significantly between time points (i.e., baseline, PA1, & PA2). Post-hoc analysis using the Bonferroni correction (3 comparison conducted) revealed that the magnitude of SOL muscle activity significantly increased (P = 0.01) from baseline to the early post-adaptation period. The magnitude of SOL (P = 0.01), MH (P = 0.04), and ABD (P = 0.049) significantly increased from baseline to the late post-adaptation period. For the pelvis assistance condition, a one-way ANOVA with repeated measures did not reveal any significant difference in the activation of all muscles between the time points (F > 0.91, P > 0.41).

Changes in SOL (t = −2.94, P = 0.02), MH (t = −2.91, P = 0.02), and ABD (t = −2.68, P = 0.02) muscle activity from baseline to the early post-adaptation period were significantly different between the two conditions and the direction of those changes was opposite. In addition, changes in SOL (t = −3.85, P = 0.003), MH (t = −2.60, P = 0.03), and ABD (t = −2.98, P = 0.01) muscle activity from baseline to the late post-adaptation period were significantly different between the two conditions, with greater improvements were observed for the resistance condition than for the assistance condition.

3.3. Overground walking after treadmill walking

A two-way ANOVA with repeated measures indicated that the condition × time interaction (F = 2.01, P = 0.16) and the main effect of condition (F = 2.31, P = 0.16) were not significant for overground walking speed. There was significant main effect of time for overground walking speed (F = 5.63, P = 0.01).

A one-way ANOVA with repeated measures revealed that overground walking speed differed significantly between time points (i.e., baseline, PA1, & PA2) for the resistance (F =6.51, P = 0.01) and assistance (F = 4.44, P = 0.03) conditions. Post-hoc analysis using the Bonferroni correction (3 comparisons conducted) showed that both experimental conditions elicited an increase in overground walking speed from the baseline to the time point 10 min post treadmill walking (resistance: P = 0.004; assistance: P = 0.048).

Changes in overground walking speed from baseline to the time point 10 min post treadmill walking was not significantly different between the two conditions (t = 0.41, P = 0.69).

The group average of stance time and its symmetry index is shown in Figure 7. A two-way ANOVA with repeated measures indicated that the condition × time interaction (F = 3.89, P = 0.04) and the main effect of condition (F = 8.86, P = 0.01) and time (F = 7.4, P = 0.004) were significant for stance time symmetry during overground walking.

A one-way ANOVA with repeated measures revealed that stance time symmetry differed significantly between time points (i.e., baseline, PA1, & PA2) for the resistance condition (F = 9.73, P = 0.001) but not for the assistance condition (F = 0.96, P = 0.40). Post-hoc analysis using the Bonferroni correction (3 comparisons conducted) showed that the resistance condition elicited an increase in stance time symmetry from the baseline to the time point 10 min post treadmill walking (P = 0.001).

All participants exhibited longer stance time for the non-paretic leg (1.11 ± 0.16) than that for the paretic leg (0.94 ± 0.17) during the baseline overground walking (t = −7.91, P = 0.00001). Changes in stance time for the paretic leg from the baseline to the time point 10 min post treadmill walking was significantly different between the two conditions (t = −3.85, P = 0.003), but there was no significant difference in the changes in stance time for the non-paretic leg between the conditions (t = −0.65, P = 0.53).

4. DISCUSSION

In this study, we tested if laterally pulling the pelvis during walking could improve use of the paretic leg and weight shift toward the paretic side in people post-stroke. Two interventions were compared: 1) Resistance Condition, which pulled the pelvis away from the paretic side during the stance phase of the paretic leg to resist weight shift, and 2) Assistance Condition, which pulled the pelvis toward the paretic side during the paretic-leg stance to assist weight shift. We found that treadmill walking with “lateral pelvis resistance” induced greater improvements in muscle activities of paretic hip abductors and ankle plantarflexors, and in lateral weight shift toward the paretic side during the post-adaptation period, compared with the effect of the “lateral pelvis assistance”. Interestingly, the improvement in lateral weight shift was associated with the improvement in muscle activity of paretic hip abductors and adductors. Further, participants exhibited increased overground walking speed 10 minutes after the treadmill walking with “lateral resistance” although the effect was comparable to the “lateral assistance” condition. These findings provide novel evidence that the application of “lateral resistance force” is effective in enhancing the use of the paretic leg and improving lateral weight shift toward the paretic side in individuals with hemiparetic stroke.

4.1. Applying lateral pelvis resistance during walking may induce improved weight shift toward the paretic side.

We found that participants showed more improvement in weight shift toward the paretic side for the pelvis resistance condition than for the pelvis assistance condition. For the pelvis assistance condition, pulling the pelvis toward the paretic side during the stance phase of the paretic leg may facilitate weight shift toward the paretic side during the early adaptation period, which was consistent with the results of a previous study (Hsu et al. 2017). However, the improved weight shift toward the paretic side induced by the pelvis assistance force might also make the participants feel that they are losing lateral balance (Watanabe 2005). Thus, the CNS might recruit additional muscle activation from the paretic leg to counteract the pelvis assistance force applied. Over multiple steps, the improved weight shift toward the paretic side gradually faded out during the mid-to-late adaptation period and even showed a slight decrease in weight shift toward the paretic side, which might be due to the overcompensation of the CNS. Following the release of the pelvis assistance force, participants showed an aftereffect consisted of further reduced weight shift toward the paretic side during the post-adaptation period, suggesting that an internal model might be formed through a feedforward control strategy to counteract the lateral pelvis assistance force (Shadmehr and Mussa-Ivaldi 1994)

In contrast, for the pelvis resistance condition, the direction of the pulling force was toward the non-paretic side during the early to mid-stance phase of the paretic leg, which was incongruent with the direction the pelvis lateral movement during the period of time. Thus, the lateral pelvis resistance force might hinder an effective weight shift toward the paretic side. As a consequence, participants showed reduced weight shift toward the paretic side during the early adaptation period, see Figure 3 and Figure 5A. However, insufficient weight shift toward the paretic side might also cause difficulty in initiating leg swing of the non-paretic leg. Thus, the CNS might recruit additional muscle activation from the paretic leg to counteract the pelvis resistance force (i.e., the resistance force might elicit additional participants’ effort to shift body weight toward the paretic side), and overtime, participants were able to generate enough counteracting force to overcome the pelvis resistance force and the weight shift toward the paretic side returned to a level that was comparable to baseline and even further toward the paretic side, which might be due to overcompensation, during the mid-to-late adaptation period. In addition, repeated exposure to the pelvis resistance force also induced motor adaptation, which was demonstrated by an aftereffect consisted of improved weight shift toward the paretic side following the release of the pelvis resistance force during the early post-adaptation period. This response is consistent with previous studies, which showed an improved symmetry of step length following the release of resistance force applied to the paretic leg (Yen et al. 2015) or the split-belt perturbation (Reisman et al. 2007).

The retention of the improved weight shift toward the paretic side during the late post-adaptation period suggests that a use-dependent motor learning mechanism might also be involved during the motor adaptation to the pelvis resistance (Diedrichsen et al. 2010). Specifically, the improvement in weight shift toward the paretic side during the late post-adaptation period was correlated with the number of paretic-leg steps with enhanced weight shifts exceeding baseline during the later adaptation period. This suggests that those participants who experienced more steps with improved weight shift toward the paretic side during the mid-to-late adaptation period also showed more improvements in weight shift toward the paretic side during the late post-adaptation period. Thus, the pelvis movement with improved weight shift toward the paretic side during the adaptation period might have contributions to the learning of improved weight shift toward the paretic side during the late post-adaptation period (Krakauer and Mazzoni 2011). This is consistent with previous studies, which suggest that repeating a particular movement with an external perturbation may change patterns of movements or muscle activation, and such behavioral or neural changes may remain even when the perturbation is removed (Classen et al. 1998). There is evidence showing that repeating reaching movements requiring to avoid an obstacle may lead to curved movement trajectories, which may be retained after the elimination of the obstacle (Jax and Rosenbaum 2007).

Thus, two fundamentally different motor learning mechanisms, i.e., the error-based motor learning and use-dependent motor learning, might act simultaneously during the motor adaptation process and provide contributions to the learning of the same motor task (Diedrichsen et al. 2010), i.e., weight shift toward the paretic side in the pelvis resistance condition, although different brain areas or neural circuities may be involved in these two motor learning mechanisms. For instance, while error-based learning may depend on the integrity of the cerebellum (Diedrichsen et al. 2005; Smith and Shadmehr 2005; Morton and Bastian 2006), use-dependent learning may depend on local changes in cortical motor area, such as primary motor cortex (Classen et al. 1998; Butefisch et al. 2000; Krakauer and Mazzoni 2011). This might be the reason why we observed different timescales of retention of improved weight shift toward the paretic side during the post-adaptation period. Error-based learning may induce fast motor adaptation, but adaptation after-effects might also be short-lived (Smith et al. 2006; Reisman et al. 2007; Shadmehr et al. 2010; Yen et al. 2015). Following the decay of the after-effect, the motor skills induced by use-dependent learning might start to emerge (Diedrichsen et al. 2010), which might be the reason participants demonstrated improved weight shift toward the paretic side during the late post-adaptation period in the pelvis resistance condition.

The improved weight shift toward the paretic side may elicit enhanced activation of the paretic leg muscles, particularly hip extensors and ankle plantarflexors. For instance, in the pelvis resistance condition, participants showed enhanced hip ABD, MH, and SOL muscle activation during the post-adaptation period. The change in muscle activation of hip ABD and ADD, key muscles primarily contributing to the control of mediolateral weight shift during walking (Winter et al. 1993; Neumann 2010; John et al. 2012) explained about 70% of the change in weight shift toward the paretic side. The enhanced activation of those muscles could improve control of mechanical impedance of hip joints (Hogan 1984) and consequently help stabilize the paretic hip joint while shifting body weight toward the paretic side. Additionally, enhanced muscle activation of MH might relate to improved mediolateral weight shift toward the paretic side because hamstring contractions contribute to stabilization of the knee joint during stance phase of the paretic leg. On the other hand, improved weight shift toward the paretic side might also require the participants to generate additional propulsion force from the paretic leg to move the body forward (Hsiao et al. 2017; Roelker et al. 2019). This might be why we observed an enhanced muscle activation of the MH (i.e., hip extensor) and SOL of the paretic leg, contributing to improved forward propulsion from the paretic leg (Roelker et al. 2019), during the post-adaptation period after the removal of the pelvis resistance force. Thus, these results also suggest that repeated exposure to the pelvis resistance during treadmill walking may induce learning of enhanced use of the paretic leg of individuals post-stroke.

4.2. Transfer of improved paretic-leg motor control from treadmill to overground

Enhanced use of the paretic leg and improved weight shift toward the paretic side elicited by treadmill walking with the application of lateral pelvis resistance might be transferred to overground walking. In this study, participants showed an increase in overground walking speed and an improvement in stance time symmetry 10 minutes after the treadmill walking for the pelvis resistance condition. This finding is parallel with findings from previous studies showing that motor adaptation achieved from treadmill walking can be transferred to overground walking in individuals with hemiparetic stroke (Reisman et al. 2009; Savin et al. 2014; Alcantara et al. 2018; Park et al. 2020). It is suggested that shared or overlapped neural circuits controlling locomotion in different contexts (e.g., treadmill vs. overground) may exist and therefore enable transfer of adapted motor control from one context to the other (Morton and Bastian 2004; Savin and Morton 2008). However, only partial motor skills may be transferred from treadmill to overground walking because environmental contexts differ for treadmill and overground walking. For instance, while participants walked on a treadmill, the treadmill belt was moving automatically so they moved their lower limbs but did not actually move forward. Also, all participants were wearing a harness and holding on the front handrail for safety during treadmill walking, which could influence gait performance (van der Veen et al. 2020). Those environmental differences between treadmill and overground walking might allow only partial transfer of adapted motor control from treadmill to overground walking (Gandevia et al. 2002; Ribeiro et al. 2013). On the other hand, while participants also showed improvement in overground walking speed after treadmill walking in the pelvis assistance condition, but showed no change in stance time symmetry, which is in contrast to the pelvis resistance condition, suggesting that participants might use different strategies, such as rely more on their non-paretic side, to improve their walking speed. This is also consistent with their motor performance, i.e., the muscle activation of the paretic leg and peak pelvis movement toward the paretic side, during treadmill walking in the pelvis assistance condition. In addition, we observed more improvements in walking speed and stance time symmetry at 10 minutes after treadmill walking rather than immediately after treadmill walking. It is possible that the immediate aftereffect might be obscured possibly by fatigue from treadmill walking. This is consistent with our recent study, which shows that an improvement in symmetrical gait patterns achieved from the treadmill walking was observed during overground walking after a short break for 10 minutes (Park et al. 2020).

Findings from this study may have potential applications for gait rehabilitation in people with chronic hemiparesis resulting from stroke. In clinics, application of manual lateral pelvis assistance by a physical therapist during treadmill walking can be used to facilitate weight shift toward the paretic side in individuals post-stroke (Hornby et al. 2008). Results from this study suggest that applying repeated lateral pelvis assistance during walking is actually less effective in inducing motor learning in weight shift toward the paretic side due to motor adaptation to the lateral assistance force. While we acknowledge that this study is only a single session of treadmill walking, the findings from this study may provide useful information for physical therapist to develop more effective intervention approaches to improve weight shift toward the paretic side in individuals post-stroke.

4.4. Limitations

In this study, all participants were allowed to hold onto the front handrail for the purpose of safety and therefore they might use their non-paretic arm to counteract the perturbation force to maintain balance. We acknowledge that during the late post-adaptation period, fatigue may increase the use of the handrails, reducing muscle activity and could impact the weight transfer to the paretic side. However, the muscle activity of the paretic leg and improved weight shift toward the paretic side during the late post-adaptation period were still significantly greater than baseline for the resistance condition. In addition, participants were allowed to hold onto the handrail for both testing conditions. Thus, we do not believe that using the handrail systematically impacted our results. Additional studies are needed to identify how participants compensate for the perturbation force by measuring force applied by the non-paretic arm using a load cell. In addition, we were not able to measure ground reaction force generated by the legs during treadmill walking due to technical limitation of the treadmill. Therefore, further studies with an instrumented treadmill are needed to determine, to what extent, participants enhance the shift of the center of pressure toward the paretic side when the pelvis resistance or assistance force is applied. We were not able to measure weight shift toward the paretic side during overground walking due to the technical limitation. In addition, we tested only short-term changes in muscle activation and motor performance. Additional studies are needed to determine the effect of the longer-term training using the same paradigm used in this study on functional walking performance.

5. CONCLUSION

Repeatedly applying lateral pelvis resistance force (i.e., pulling the pelvis toward the non-paretic side) rather than lateral pelvis assistance force toward the paretic side during the stance phase of the paretic leg may be more effective in inducing learning of improved lateral weight shift toward the paretic side and enhanced use of the paretic leg in hemiparetic stroke survivors. Further, enhanced use of the paretic leg and improved weight shift toward the paretic side may transfer from treadmill to overground walking, resulting in improved walking speed and stance time symmetry. Results from this study also suggest that two fundamentally different motor learning mechanisms, i.e., the error-based motor learning and use-dependent motor learning, may act simultaneously and provide contributions to the learning of improved weight shift toward the paretic side in individuals post-stroke. Knowledge gained from this study provides insights into development of locomotor training protocols in order to improve weight shift toward the paretic side in individuals post-stroke.

Acknowledgements:

This work was supported by the National Institute of Health (R01HD082216). No potential conflict of interest was reported by the authors.

Footnotes

Code availability: Customized LabVIEW and MATLAB codes will not be shared for intellectual property protection.

Availability of data and material:

Data can be provided at request.

REFERENCES

Uncategorized References

  1. Alcantara CC, Charalambous CC, Morton SM, Russo TL, Reisman DS (2018) Different Error Size During Locomotor Adaptation Affects Transfer to Overground Walking Poststroke. Neurorehabil Neural Repair 32:1020–1030 doi: 10.1177/1545968318809921 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Almkvist Muren M, Hütler M, Hooper J (2008) Functional capacity and health-related quality of life in individuals post stroke. Topics in stroke rehabilitation 15:51–58 [DOI] [PubMed] [Google Scholar]
  3. Ardestani MM, Kinnaird CR, Henderson CE, Hornby TG (2019) Compensation or Recovery? Altered Kinetics and Neuromuscular Synergies Following High-Intensity Stepping Training Poststroke. Neurorehabil Neural Repair 33:47–58 doi: 10.1177/1545968318817825 [DOI] [PubMed] [Google Scholar]
  4. Barbeau H (2003) Locomotor training in neurorehabilitation: emerging rehabilitation concepts. Neurorehabilitation and Neural Repair 17:3–11 [DOI] [PubMed] [Google Scholar]
  5. Bohannon RW, Andrews AW, Smith MB (1988) Rehabilitation goals of patients with hemiplegia. International Journal of Rehabilitation Research 11:181–184 [Google Scholar]
  6. Butefisch CM, Davis BC, Wise SP, Sawaki L, Kopylev L, Classen J, Cohen LG (2000) Mechanisms of use-dependent plasticity in the human motor cortex. Proc Natl Acad Sci U S A 97:3661–3665 doi: 10.1073/pnas.050350297 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Classen J, Liepert J, Wise SP, Hallett M, Cohen LG (1998) Rapid plasticity of human cortical movement representation induced by practice. J Neurophysiol 79:1117–1123 doi: 10.1152/jn.1998.79.2.1117 [DOI] [PubMed] [Google Scholar]
  8. Diedrichsen J, Verstynen T, Lehman SL, Ivry RB (2005) Cerebellar involvement in anticipating the consequences of self-produced actions during bimanual movements. J Neurophysiol 93:801–812 doi: 10.1152/jn.00662.2004 [DOI] [PubMed] [Google Scholar]
  9. Diedrichsen J, White O, Newman D, Lally N (2010) Use-dependent and error-based learning of motor behaviors. J Neurosci 30:5159–5166 doi: 10.1523/JNEUROSCI.5406-09.2010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Folstein MF, Robins LN, Helzer JE (1983) The mini-mental state examination. Archives of general psychiatry 40:812–812 [DOI] [PubMed] [Google Scholar]
  11. Gandevia SC, Refshauge KM, Collins DF (2002) Proprioception: peripheral inputs and perceptual interactions. In: Sensorimotor control of movement and posture. Springer, pp 61–68 [DOI] [PubMed] [Google Scholar]
  12. Gordon CD, Wilks R, McCaw-Binns A (2013) Effect of aerobic exercise (walking) training on functional status and health-related quality of life in chronic stroke survivors: a randomized controlled trial. Stroke 44:1179–1181 [DOI] [PubMed] [Google Scholar]
  13. Grieve D, Gear RJ (1966) The relationships between length of stride, step frequency, time of swing and speed of walking for children and adults. Ergonomics 9:379–399 [DOI] [PubMed] [Google Scholar]
  14. Hogan N (1984) Adaptive control of mechanical impedance by coacti- vation of antagonist muscles. IEEE Trans Auto Control 29:9 [Google Scholar]
  15. Hornby TG, Campbell DD, Kahn JH, Demott T, Moore JL, Roth HR (2008) Enhanced gait-related improvements after therapist- versus robotic-assisted locomotor training in subjects with chronic stroke: a randomized controlled study. Stroke 39:1786–1792 doi: 10.1161/STROKEAHA.107.504779 [DOI] [PubMed] [Google Scholar]
  16. Houldin A, Luttin K, Lam T (2011) Locomotor adaptations and aftereffects to resistance during walking in individuals with spinal cord injury. J Neurophysiol 106:247–258 doi: 10.1152/jn.00753.2010 [DOI] [PubMed] [Google Scholar]
  17. Hsiao H, Gray VL, Creath RA, Binder-Macleod SA, Rogers MW (2017) Control of lateral weight transfer is associated with walking speed in individuals post-stroke. Journal of Biomechanics 60:72–78 doi: 10.1016/j.jbiomech.2017.06.021 [DOI] [PubMed] [Google Scholar]
  18. Hsu CJ, Kim J, Roth EJ, Rymer WZ, Wu M (2019) Use of Pelvic Corrective Force With Visual Feedback Improves Paretic Leg Muscle Activities and Gait Performance After Stroke. IEEE Trans Neural Syst Rehabil Eng 27:2353–2360 doi: 10.1109/tnsre.2019.2950226 [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Hsu CJ, Kim J, Tang R, Roth EJ, Rymer WZ, Wu M (2017) Applying a pelvic corrective force induces forced use of the paretic leg and improves paretic leg EMG activities of individuals post-stroke during treadmill walking. Clin Neurophysiol 128:1915–1922 doi: 10.1016/j.clinph.2017.07.409 [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Jax SA, Rosenbaum DA (2007) Hand path priming in manual obstacle avoidance: evidence that the dorsal stream does not only control visually guided actions in real time. J Exp Psychol Hum Percept Perform 33:425–441 doi: 10.1037/0096-1523.33.2.425 [DOI] [PubMed] [Google Scholar]
  21. John CT, Seth A, Schwartz MH, Delp SL (2012) Contributions of muscles to mediolateral ground reaction force over a range of walking speeds. Journal of biomechanics 45:2438–2443 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Kamono A, Ogihara N (2018) Weight-shift ability significantly correlates with walking velocity in post-acute stroke patients. Proc Inst Mech Eng H 232:361–370 doi: 10.1177/0954411918757814 [DOI] [PubMed] [Google Scholar]
  23. Krakauer JW, Mazzoni P (2011) Human sensorimotor learning: adaptation, skill, and beyond. Curr Opin Neurobiol 21:636–644 doi: 10.1016/j.conb.2011.06.012 [DOI] [PubMed] [Google Scholar]
  24. Lauziere S, Betschart M, Aissaoui R, Nadeau S (2014) Understanding spatial and temporal gait asymmetries in individuals post stroke. Int J Phys Med Rehabil 2:201 [Google Scholar]
  25. Malone LA, Bastian AJ (2014) Spatial and temporal asymmetries in gait predict split-belt adaptation behavior in stroke. Neurorehabil Neural Repair 28:230–240 doi: 10.1177/1545968313505912 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Mehrholz J, Thomas S, Elsner B (2017) Treadmill training and body weight support for walking after stroke. Cochrane Database Syst Rev 8:CD002840 doi: 10.1002/14651858.CD002840.pub4 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Morton SM, Bastian AJ (2004) Prism adaptation during walking generalizes to reaching and requires the cerebellum. J Neurophysiol 92:2497–2509 doi: 10.1152/jn.00129.2004 [DOI] [PubMed] [Google Scholar]
  28. Morton SM, Bastian AJ (2006) Cerebellar contributions to locomotor adaptations during splitbelt treadmill walking. J Neurosci 26:9107–9116 doi: 10.1523/JNEUROSCI.2622-06.2006 [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Nascimento LR, Boening A, Galli A, Polese JC, Ada L (2021) Treadmill walking improves walking speed and distance in ambulatory people after stroke and is not inferior to overground walking: a systematic review. J Physiother 67:95–104 doi: 10.1016/j.jphys.2021.02.014 [DOI] [PubMed] [Google Scholar]
  30. Neumann DA (2010) Kinesiology of the hip: a focus on muscular actions. journal of orthopaedic & sports physical therapy 40:82–94 [DOI] [PubMed] [Google Scholar]
  31. Nindorera F, Nduwimana I, Thonnard JL, Kossi O (2021) Effectiveness of walking training on balance, motor functions, activity, participation and quality of life in people with chronic stroke: a systematic review with meta-analysis and meta-regression of recent randomized controlled trials. Disabil Rehabil:1–12 doi: 10.1080/09638288.2021.1894247 [DOI] [PubMed] [Google Scholar]
  32. O’lney SJ, Richards C (1996) Hemiparetic gait following stroke. Part I: Characteristics. Gait & Posture 4:13 doi: 10.1016/0966-6362(96)01063-6 [DOI] [Google Scholar]
  33. Park SH, Lin J-T, Dee W, Hsu C-J, Roth EJ, Rymer WZ, Wu M (2020) Targeted pelvic constraint force induces enhanced use of the paretic leg during walking in persons post-stroke. IEEE Transactions on Neural Systems and Rehabilitation Engineering [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Patton JL, Stoykov ME, Kovic M, Mussa-Ivaldi FA (2006) Evaluation of robotic training forces that either enhance or reduce error in chronic hemiparetic stroke survivors. Exp Brain Res 168:368–383 doi: 10.1007/s00221-005-0097-8 [DOI] [PubMed] [Google Scholar]
  35. Plummer P, Behrman AL, Duncan PW, et al. (2007) Effects of stroke severity and training duration on locomotor recovery after stroke: a pilot study. Neurorehabilitation and neural repair 21:137–151 [DOI] [PubMed] [Google Scholar]
  36. Raja B, Neptune RR, Kautz SA (2012) Coordination of the non-paretic leg during hemiparetic gait: expected and novel compensatory patterns. Clinical Biomechanics 27:1023–1030 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Reisman DS, Wityk R, Silver K, Bastian AJ (2007) Locomotor adaptation on a split-belt treadmill can improve walking symmetry post-stroke. Brain 130:1861–1872 doi: 10.1093/brain/awm035 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Reisman DS, Wityk R, Silver K, Bastian AJ (2009) Split-belt treadmill adaptation transfers to overground walking in persons poststroke. Neurorehabil Neural Repair 23:735–744 doi: 10.1177/1545968309332880 [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Ribeiro T, Britto H, Oliveira D, Silva E, Galvao E, Lindquist A (2013) Effects of treadmill training with partial body weight support and the proprioceptive neuromuscular facilitation method on hemiparetic gait: a randomized controlled study. Eur J Phys Rehabil Med 49:451–461 [PubMed] [Google Scholar]
  40. Roelker SA, Bowden MG, Kautz SA, Neptune RR (2019) Paretic propulsion as a measure of walking performance and functional motor recovery post-stroke: a review. Gait & posture 68:6–14 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Roerdink M, Geurts AC, de Haart M, Beek PJ (2009) On the relative contribution of the paretic leg to the control of posture after stroke. Neurorehabilitation and neural repair 23:267–274 [DOI] [PubMed] [Google Scholar]
  42. Savin DN, Morton SM (2008) Asymmetric generalization between the arm and leg following prism-induced visuomotor adaptation. Exp Brain Res 186:175–182 doi: 10.1007/s00221-007-1220-9 [DOI] [PubMed] [Google Scholar]
  43. Savin DN, Morton SM, Whitall J (2014) Generalization of improved step length symmetry from treadmill to overground walking in persons with stroke and hemiparesis. Clin Neurophysiol 125:1012–1020 doi: 10.1016/j.clinph.2013.10.044 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Shadmehr R, Mussa-Ivaldi FA (1994) Adaptive representation of dynamics during learning of a motor task. J Neurosci 14:3208–3224 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Shadmehr R, Smith MA, Krakauer JW (2010) Error correction, sensory prediction, and adaptation in motor control. Annu Rev Neurosci 33:89–108 doi: 10.1146/annurev-neuro-060909-153135 [DOI] [PubMed] [Google Scholar]
  46. Smith MA, Ghazizadeh A, Shadmehr R (2006) Interacting adaptive processes with different timescales underlie short-term motor learning. PLoS Biol 4:e179 doi: 10.1371/journal.pbio.0040179 [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Smith MA, Shadmehr R (2005) Intact ability to learn internal models of arm dynamics in Huntington’s disease but not cerebellar degeneration. J Neurophysiol 93:2809–2821 doi: 10.1152/jn.00943.2004 [DOI] [PubMed] [Google Scholar]
  48. Taub E, Crago JE, Burgio LD, Groomes TE, Cook EW 3rd, DeLuca SC, Miller NE (1994) An operant approach to rehabilitation medicine: overcoming learned nonuse by shaping. J Exp Anal Behav 61:281–293 doi: 10.1901/jeab.1994.61-281 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Taub E, Uswatte G, Bowman MH, et al. (2013a) Constraint-induced movement therapy combined with conventional neurorehabilitation techniques in chronic stroke patients with plegic hands: a case series. Arch Phys Med Rehabil 94:86–94 doi: 10.1016/j.apmr.2012.07.029 [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Taub E, Uswatte G, King DK, Morris D, Crago JE, Chatterjee A (2006) A placebo-controlled trial of constraint-induced movement therapy for upper extremity after stroke. Stroke 37:1045–1049 [DOI] [PubMed] [Google Scholar]
  51. Taub E, Uswatte G, Mark VW, et al. (2013b) Method for enhancing real-world use of a more affected arm in chronic stroke: transfer package of constraint-induced movement therapy. Stroke 44:1383–1388 doi: 10.1161/STROKEAHA.111.000559 [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. van der Veen SM, Hammerbeck U, Hollands KL (2020) How accuracy of foot-placement is affected by the size of the base of support and crutch support in stroke survivors and healthy adults. Gait & Posture 76:224–230 [DOI] [PubMed] [Google Scholar]
  53. Watanabe Y (2005) Fear of falling among stroke survivors after discharge from inpatient rehabilitation. International Journal of Rehabilitation Research 28:149–152 [DOI] [PubMed] [Google Scholar]
  54. Winter DA, MacKinnon CD, Ruder GK, Wieman C (1993) An integrated EMG/biomechanical model of upper body balance and posture during human gait. Prog Brain Res 97:359–367 doi: 10.1016/s0079-6123(08)62295-5 [DOI] [PubMed] [Google Scholar]
  55. Wolf SL, Winstein CJ, Miller JP, et al. (2008) Retention of upper limb function in stroke survivors who have received constraint-induced movement therapy: the EXCITE randomised trial. Lancet Neurol 7:33–40 doi: 10.1016/S1474-4422(07)70294-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Wu M, Hornby TG, Landry JM, Roth H, Schmit BD (2011) A cable-driven locomotor training system for restoration of gait in human SCI. Gait Posture 33:256–260 doi: 10.1016/j.gaitpost.2010.11.016 [DOI] [PubMed] [Google Scholar]
  57. Wu M, Hsu CJ, Kim J (2017) Motor adaptation to lateral pelvis assistance force during treadmill walking in individuals post-stroke. IEEE Int Conf Rehabil Robot 2017:300–303 doi: 10.1109/ICORR.2017.8009263 [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Wu M, Hsu CJ, Kim J (2019) Forced use of paretic leg induced by constraining the non-paretic leg leads to motor learning in individuals post-stroke. Exp Brain Res 237:2691–2703 doi: 10.1007/s00221-019-05624-w [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Yen S-C, Schmit BD, Landry JM, Roth H, Wu M (2012) Locomotor adaptation to resistance during treadmill training transfers to overground walking in human SCI. Experimental brain research 216:473–482 [DOI] [PubMed] [Google Scholar]
  60. Yen SC, Landry JM, Wu M (2013) Size of kinematic error affects retention of locomotor adaptation in human spinal cord injury. J Rehabil Res Dev 50:1187–1200 doi: 10.1682/JRRD.2012.09.0175 [DOI] [PubMed] [Google Scholar]
  61. Yen SC, Schmit BD, Wu M (2015) Using swing resistance and assistance to improve gait symmetry in individuals post-stroke. Hum Mov Sci 42:212–224 doi: 10.1016/j.humov.2015.05.010 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

Data can be provided at request.

RESOURCES