Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Kinetic Gait Changes after Robotic Exoskeleton Training in Adolescents and Young Adults with Acquired Brain Injury

Kinetic Gait Changes after Robotic Exoskeleton Training in Adolescents and Young Adults with... Hindawi Applied Bionics and Biomechanics Volume 2020, Article ID 8845772, 10 pages https://doi.org/10.1155/2020/8845772 Research Article Kinetic Gait Changes after Robotic Exoskeleton Training in Adolescents and Young Adults with Acquired Brain Injury 1,2,3 1,3 2,3 Kiran K. Karunakaran , Naphtaly Ehrenberg , JenFu Cheng , 2,3 1,2,3 Katherine Bentley, and Karen J. Nolan Center for Mobility and Rehabilitation Engineering Research, Kessler Foundation, West Orange, New Jersey 07052, USA Physical Medicine and Rehabilitation, Rutgers New Jersey Medical School, Newark, New Jersey 07103, USA Research Department, Children's Specialized Hospital, New Brunswick, New Jersey 08901, USA Correspondence should be addressed to Kiran K. Karunakaran; kkarunakaran@kesslerfoundation.org Received 15 July 2020; Revised 11 September 2020; Accepted 8 October 2020; Published 28 October 2020 Academic Editor: Dongming Gan Copyright © 2020 Kiran K. Karunakaran et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Background. Acquired brain injury (ABI) is one of the leading causes of motor deficits in children and adults and often results in motor control and balance impairments. Motor deficits include abnormal loading and unloading, increased double support time, decreased walking speed, control, and coordination. These deficits lead to diminished functional ambulation and reduced quality of life. Robotic exoskeletons (RE) for motor rehabilitation can provide the user with consistent, symmetrical, goal-directed repetition of movement, as well as balance and stability. Purpose. The goal of this preliminary prospective before and after study is to evaluate the therapeutic effect of RE training on the loading/unloading and spatial-temporal characteristics in adolescents and young adults with chronic ABI. Method. Seven participants diagnosed with ABI between the ages of 14 and 27 years participated in the study. All participants received twelve 45 minute sessions of RE gait training. The bilateral loading (linearity of loading and rate of loading), speed, step length, swing time, stance time, and total time were collected using Zeno™ walkway (ProtoKinetics, Havertown, PA, USA) before and after RE training. Results. Results from the study showed improved step length, speed, and an overall progression towards healthy bilateral loading, with linearity of loading showing a significant therapeutic effect (p <0:05). Conclusion. These preliminary results suggest that high dose, repetitive, consistent gait training using RE has the potential to induce recovery of function in adolescents and young adults diagnosed with ABI. 1. Introduction needed for a functional recovery [10]. Consequently, after motor rehabilitation individuals with ABI can experience Acquired brain injury (ABI) is a leading cause of hemiparesis variable recovery with residual gait deviations. As a result resulting in gait and balance deficits in adolescents and young of these residual motor and balance deficits, individuals diag- adults [1–3]. These deficits result in abnormal loading and nosed with ABI may develop compensatory mechanisms unloading, increased double support time, and decreased such as abnormal or asymmetrical loading and unloading walking speed, control, and coordination, leading to impair- characteristics and prolonged weight transfer, in order to ment in functional ambulation and associated activities of achieve ambulation [4, 11]. daily living [2–5]. Recovery involves rehabilitation through Ground reaction force (GRF) parameters, e.g., peak high dose physical therapy [6]. It is based on the theory that values, linearity, and rates of loading/unloading, symmetry the human brain is capable of self-reorganization, or plastic- coefficients, and force integrals, are used for assessment of pathological gait and evaluation of therapeutic efficacy ity, through continuous, consistent, repeated practice aimed at restoring function and independence [7–9]. Conventional [12–14]. The analysis of vertical forces during gait provides therapy alone may not be able to provide enough consistent information on weight bearing and balance, and these force mass practice and repetition to facilitate the neuroplasticity or pressure patterns in the paretic leg are known to be 2 Applied Bionics and Biomechanics Table 1: Subject demographics. Subject Condition Affected side Gender Age Height (m) Weight (Kg) Years since injury 1 TBI Left Male 23 1.83 90.00 1 2 Stroke Left Female 23 1.60 54.00 .9 3 Anoxic Left Female 17 1.63 54.00 2 4 Stroke Left Male 16 1.83 67.50 .5 5 TBI Left Male 22 1.78 63.50 3 6 TBI Left Female 27 1.63 62.60 5 7 TBI Right Male 14 1.65 49.44 0.5 Mean ± SD 20:29 ± 4:68 1:71 ± 0:10 63:01 ± 13:51 1:84 ± 1:66 1 HC n/a Male 26 1.57 77.11 n/a of the RE gait training environment for reference, to help correlated with walking speed and motor recovery in stroke patients [15, 16]. Therefore, improvements in vertical pres- understand the therapeutic effects. sure profiles might demonstrate improvement in gait and overall functional ambulation. 2. Materials and Methods In a healthy gait cycle, the loading response is bilaterally symmetrical [17]. It has three distinct phases, starting with This is a preliminary prospective before and after study a peak force during the initial loading phase, followed by a which is to evaluate the therapeutic effect of RE training on midstance phase where there is a decrease in loading, and the loading/unloading and spatial-temporal characteristics ending with another peak force during the terminal unload- in adolescents and young adults with chronic ABI. ing phase [17]. Thus, the loading response resembles a bimodal M shape [17]. The transfer of weight during the 2.1. Participants. Twelve participants with ABI were initial loading phase allows for the efficient transfer of recruited for this study. Kinetic and temporal-spatial data momentum from one leg to the other and the uninter- for baseline and follow-up visits were only available from rupted use of the kinetic energy that is created by swing nine participants. Further, two participants were excluded limb activity [18]. The rate of loading determines the speed from the analysis as they were diagnosed with bilateral defi- of gait [15]. Research has shown that smooth linear loading cits. Therefore, seven participants diagnosed with ABI and helps conserve momentum and shifts the body’s weight to hemiparesis and a single healthy control (HC) were included the next phase of the gait cycle while maintaining speed for this analysis (Table 1). The inclusion criteria for ABI par- [14, 17, 19]. If weight transfer is delayed or loading is non- ticipants in this investigation included (1) diagnosed with an linear, it will result in a loss of energy and hence less effi- acquired brain injury (anoxic, stroke, or TBI); (2) between cient gait, leading to decreased speed and necessitating the the ages of 13 and 28; (3) able to walk with or without an use of other compensatory gait mechanisms. Limb loading assistive device; (4) no additional orthopedic, neuromuscu- is often inefficient in individuals with ABI because the lar, or severe neurological pathologies (unrelated to their affected limb has a difficult time accepting weight, resulting ABI) that would interfere with their ability to walk; (5) able in a nonlinear loading and decreased momentum. Conse- to stand upright for 30 minutes with assistance; and (6) able quently, a targeted goal during gait rehabilitation in ABI to physically fit into the RE (weight ≤ 100 kg, height ≤ 1.88 m, is to improve loading profiles in order to facilitate healthier hip width 0.36 m-0.46 m). The inclusion criteria for the HC movement patterns. were no orthopedic, neuromuscular, or severe neurological Lower extremity robotic exoskeletons (REs) are currently pathologies that would interfere with gait and balance. All being used in rehabilitation to restore gait functionality. REs procedures performed in this investigation were approved have the potential to provide the user with high dose, consis- by the Human Subjects Institutional Review Board at Kessler tent, symmetrical bilateral loading [9, 20–26] profiles during Foundation, and informed consent was obtained prior to gait. It can provide an increased number of steps (increased study participation. dosing) in a consistent and controlled training environment, which is ideal for inducing neuroplasticity [20, 27]. This is 2.2. Robotic Exoskeleton Gait Training. Robotic gait training essential for chronic ABI patients who need high dose, con- in bilateral assistance mode was administered as an outpa- sistent therapy to induce cortical reorganization for func- tient rehabilitation gait intervention at the Kessler Founda- tional recovery. This study utilized an RE to provide tion using a commercially available robotic exoskeleton intensive gait training to adolescents and young adults diag- (Figure 1, EksoGT, Ekso Bionics, Inc., Richmond, CA, nosed with an ABI, with the goal of evaluating the efficacy of USA) for 45 minutes per day for 12 days over a period of 4 high repetition robotic training on loading/unloading pro- weeks. A licensed physical therapist administered all RE gait files. The objective of this preliminary prospective study training sessions and adjusted the assistance provided by the was to evaluate the therapeutic effect of RE on their loadin- robot to the individual participant’s therapy progression. A g/unloading characteristics. This paper also provides details member of the study team was present at all times during Applied Bionics and Biomechanics 3 Figure 1: Robotic gait training with a participant with ABI administered by a trained physical therapist. lateral weight shift to a predetermined distance from mid- the RE gait training sessions to assist the physical therapist and ensure participant safety. The healthy control participant line to initiate advancement of the back (preswing) leg. was not given RE training as part of the investigation. The RE provided overground gait rehabilitation under 2.3. Data Collection Procedures. The participants with ABI the guidance of a licensed physical therapist. The RE’s participated in two data collection sessions (baseline and upper section is attached to the user’s upper body, with follow-up after 12 sessions of outpatient RE training), and a backpack style shoulder harness and torso brace and also data was collected while walking with (training environment) houses the battery pack, while the lower sections are and without the RE. The participants with ABI did not utilize affixed to the legs with upper thigh straps, shin guards, any orthotic devices or dorsiflexion wrap during data collec- and secure foot bindings. The RE has two active degrees tion. The reference HC participated in one data collection of freedom at the hip and the knee, respectively, and a session, and data was collected while walking without the RE. passively sprung ankle joint with adjustable stiffness in During each data collection session, temporal, spatial, the sagittal plane. The RE provides assistive torques to and loading data was collected using a Zeno™ walkway (Pro- the hip and knee joints to perform the predefined gait tra- toKinetics, Havertown, PA, USA) at 120 Hz. The participant jectory and provides variable assistance as required by the performed up to 6 walking trials (approximately six 10 meter participants bilaterally. The actuated range of motion at walks) at a self-selected pace per condition (with and without ° ° the hip is -20 to 135 , and the actuated range for the knee RE) and wore shoes for all walking assessments. Participants ° ° is 0 to 120 [28]. All steps were initiated by the partici- were allowed to rest or take breaks at any time during testing pant. In order to trigger a step, the subject performed a to minimize the effects of fatigue. A member of the study 4 Applied Bionics and Biomechanics Table 2: Outcome measures. Outcome measure Description Statistical analysis TVP during the stance phase of each gait cycle was computed, and an average TVP for all gait cycles was calculated for both the legs without and with RE that was computed for each subject and for the HC. The TVP was normalized to 100% of stance in each condition for comparison. The stance phase was further divided into Total vertical initial double support (IDS) and terminal double support pressure (TVP) (TDS) phases based on heel strike and toe off. The IDS pressure was computed as the pressure between ipsilateral heel strikes to contralateral toe off. The TDS pressure was computed as the pressure between contralateral heel strikes to ipsilateral toe off. Mean and standard deviation TVP for all participants with ABI was computed. A best fit line was computed for the average IDS loading phase for each session for each subject. A goodness of the fit was computed to assess the error between the fitted line and the average loading during IDS for each subject in Kolmogorov- Smirnov Z test (p <0:05) of normality Linearity of each session. showed that the data were not normal. Wilcoxson signed loading (goodness The goodness of fit was used to assess the smooth linearity rank test was used to determine the therapeutic effect of fit) of loading. R-square (R ) was computed to assess the (baseline to follow-up without RE) on goodness of fit. square of the correlation between average loading during IDS and the best fit line. A higher R value signifies a closer fit to the best fit line or increased linearity. Kolmogorov- Smirnov Z test (p >0:05) of normality The slope of the average IDS loading phase was computed Rate of linear showed that the data were normal. A paired sample for each subject. Slope indicates the rate of linear loading. loading (slope of t-test was performed to determine the therapeutic effect Increased slope in the IDS phase indicates an increased initial loading) (baseline to follow-up without RE) on the slope of initial moment during the first rocker. loading. Kolmogorov- Smirnov Z test (p > :05) of normality The average walking speed was computed for each subject showed that the data were normal. A paired sample Walking speed as the linear distance with respect to time to complete a t-test was performed to determine the therapeutic effect gait cycle. (baseline to follow-up without RE) on walking speed. The average step length for each gait cycle was computed Kolmogorov- Smirnov Z test (p > :05) of normality as the forward linear displacement between foot contact of showed that the data were normal. A paired sample Step length the ipsilateral leg to foot contact of the contralateral leg t-test was performed to determine the therapeutic effect during each gait cycle. Average step length was computed (baseline to follow-up without RE) on step length. for each subject. Total time was computed as the time between foot contact of one leg to the subsequent foot contact of the same leg. The average total time was computed for each gait cycle. Kolmogorov- Smirnov Z test (p < :05) of normality Further, average swing time for each subject during each showed that the data were not normal. Wilcoxson signed Temporal condition was computed as the time between the foot off rank test was used to the therapeutic effect (baseline to measures the floor of one leg to foot contact of the same leg during follow-up without RE) on total time, swing time, and the gait cycle. Average stance time for each subject during stance time. each condition was computed as the time between the foot contact of one leg to toe off the same leg during the gait cycle. team was present with the participants at all times during the ized pressure data were exported. Custom MATLAB algo- rithms were used to analyze the gait trajectories for each walking trials. During the baseline and follow-up gait assessments in the session with and without the RE. The exported data were RE, torque was provided bilaterally as needed at the hip and further divided into gait cycles, with a gait cycle being knee to complete the gait. defined as the period from ground contact of one foot to the subsequent ground contact of the same foot. Data for up to 15 gait cycles per condition and per session were 2.4. Data Analysis. PKMAS (ProtoKinetics Havertown, PA, USA) and MATLAB (MathWorks, Natick, MA, USA) were available for each subject and were used for data analysis. used for data analysis. The data were preprocessed using Selected outcome measures and statistical analyses are pre- the PKMAS software, and temporal, spatial, and normal- sented in Table 2. Applied Bionics and Biomechanics 5 Baseline Follow-up Healthy control Without RE Without RE Without RE 400 400 400 VP1 VP3 300 300 300 VP2 P2 P1 200 200 200 100 100 100 0 0 0 0 50 100 0 50 100 0 50 100 With RE With RE Stance phase in percentage Right Left 500 500 400 400 300 300 200 200 100 100 0 0 0 50 100 0 50 100 Stance phase in percentage Affected Unaffected Figure 2: Mean ± standard deviation of the TVP of the affected and unaffected leg of individuals with ABI during walking with and without an RE at baseline and follow-up and one reference HC. Data is normalized to 100% of the stance phase. 3. Results deviation from the mean across gait cycles for participants with ABI was smaller in the RE training environment com- 3.1. Total Vertical Pressure (TVP). Figure 2 shows the average pared to without the RE at both baseline and follow-up. TVP of all participants with ABI at baseline and at follow-up 3.2. Rate of Linear Loading (Slope of Initial Loading). There for both the affected and the unaffected sides with and was an increase in the average slope from baseline to without RE and one reference HC. The reference HC’s TVP follow-up while walking without the RE (Table 3). A paired profile demonstrated linear loading and unloading, a sym- sample t-test for average slope did not show a significant metrical loading pattern during stance bilaterally, and a min- therapeutic effect between baseline and follow-up, though imal deviation from the mean across gait cycles. For the effect size (p =0:136, Cohen’s d effect size was 0.59) was participants with ABI walking without RE at baseline, the high (Table 3). The average slope was higher in the training TVP demonstrated a perturbation during loading for the environment (with the RE) both at baseline and at follow- IDS phase (Figure 2-P1and Figure 3-P3) and increased load- up compared to without the RE. The RE training environ- ing on the unaffected side. At follow-up, the perturbation ment was similar to HC (Table 3). during loading for IDS decreased (Figure 2-P2 and Figure 3-P3), and there was increased loading on the affected 3.3. Linearity of Loading (Goodness of Fit). There was an limb during midstance. There was increased variation in the increase in average R from baseline to follow-up while walk- TVP on the affected side compared to the unaffected side for ing without the RE (Table 3). The Wilcoxson signed rank test individuals with ABI, demonstrated by the standard deviations. In the RE training environment, the TVP profile showed for R showed a significant therapeutic effect between base- line and follow-up, and the effect size (p =0:018, Cohen’s linear loading during IDS, but did not show the distinctive peaks as observed in the HC’s data (Figure 2-VP1 and d effect size was 0.6334) was high. The R of loading Figure 2-VP3) in both the affected and unaffected sides at was higher in the training environment (with the RE) at both baseline and follow-up. Furthermore, a bilaterally sym- baseline and about the same at follow-up compared to metrical loading profile was observed in the RE training envi- without the RE. The RE training environment was similar ronment and for the HC as compared to without RE. The to the HC (Table 3). Vertical pressure 6 Applied Bionics and Biomechanics Baseline Follow-up Healthy control Without RE Without RE Without RE 400 400 300 300 300 P3 P4 200 200 200 100 100 0 0 0 50 100 0 50 100 0 50 100 With RE With RE IDS phase in percentage Right Left 500 500 Best fit-right Best fit-left 400 400 300 300 200 200 100 100 0 0 0 50 100 0 50 100 IDS phase in percentage Affected Unaffected Best fit-affected Best fit-unaffected Figure 3: Mean ± standard deviation of the initial loading phase of the right and left leg of subjects with ABI while walking with and without the RE at baseline and at follow-up and a reference HC. The dotted lines are the best fit lines for the loading profile. Table 3: Mean ± standard error of initial double support (IDS) loading characteristics on the affected side of all participants with ABI and IDS loading characteristics on the left side of one HC. Metric Baseline-without RE Follow-up-without RE Baseline-with RE Follow-up-with RE HC 2:33 ± 0:28 2:60 ± 0:36 3:44 ± 0:29 3:43 ± 0:24 Slope 3.46 0:889 ± 0:05 0:934 ± 0:04 0:927 ± 0:02 0:923 ± 0:02 Goodness of fit 0.99 3.4. Step Length. Step length without the RE showed an change (Figure 2(c)). A paired sample t-test did not show a increase in both the unaffected and affected sides at follow- significant therapeutic effect between baseline and follow- up, though the effect size (p =0:083, Cohen’s d effect size up compared to baseline in subjects 2,3,5,6, and 7, while it showed a decrease in subject 1 and no change on the unaf- was 0.80) was high. The average walking speed was lower in fected side and an increase on the affected side in subject 4 the training environment (with the RE) compared to without (Figures 2(a) and 2(b)). A paired sample t-test did not show the RE at baseline and follow-up. The average speed of the a significant therapeutic effect between baseline and follow- HC data was higher than ABI without RE. up, though the effect size (p = :075, Cohen’s d effect size 3.6. Temporal Characteristics. Average total time, stance was 0.81) was high. The average step length was lower in time, and total double support time decreased from baseline the training environment (with the RE) compared to without to follow-up (Table 4), but did not show a significant the RE at baseline and follow-up. The average step length of therapeutic effect (p =0:612, p =0:398, the HC data was higher than ABI without RE. totaltime stance time p =0:237). The average total time, stance total double support time 3.5. Walking Speed. Walking speed without the RE increased time, and total double support time were higher in the train- from baseline to follow-up for subjects 2,4,5,6, and 7, while ing environment (with the RE) compared to without the RE subject 1 decreased their speed, and subject 3 showed no at baseline and at follow-up. Average total time, stance time, Vertical pressure Applied Bionics and Biomechanics 7 Table 4: Mean ± standard error of temporal characteristics. Metric Baseline-without RE Follow-up-without RE Baseline-with RE Follow-up-with RE HC Total time 2:18 ± 0:69 1:98 ± 0:50 3:78 ± 0:17 3:35 ± 0:09 1.24 1:66 ± 0:62 1:46 ± 0:45 3:12 ± 0:15 2:69 ± 0:10 Stance time 0.80 Affected 0:52 ± 0:07 0:52 ± 0:06 0:66 ± 0:05 0:66 ± 0:06 Swing time 0.44 1:25 ± 0:62 1:05 ± :045 2:40 ± 0:17 2:04 ± 0:10 Total double support 0.36 2:15 ± 0:66 1:95 ± 0:48 3:77 ± 0:16 3:33 ± 0:10 Total time 1.27 1:74 ± 0:66 1:55 ± 0:47 3:05 ± 0:17 2:69 ± 0:10 Stance time 0.85 Unaffected 0:41 ± 0:04 0:40 ± 0:04 0:72 ± 0:11 0:64 ± 0:02 Swing time 0.43 0:73 ± 0:49 0:44 ± 0:21 1:19 ± 0:10 1:01 ± 0:05 Total double support 0.23 during initial loading (Figure 2-P1, Figure 3-P3) on the and total double support time of the HC data were lower than ABI without RE. affected side during baseline, which indicates a nonlinear ini- Swing time did not change (p =0:866) from baseline to tial loading response. Linear loading directly contributes to follow-up (Table 4). The average swing time was higher in the momentum during gait [30]. Therefore, any nonlinearity the training environment (with the RE) compared to without would result in decreased momentum and slower load trans- the RE at baseline and follow-up. Average swing time of the fer during gait, leading to a disruption in forward progres- HC data was lower than ABI without RE. sion. ABI patients with hemiplegia often present with reduced hip flexion, dorsiflexion, and plantarflexion during the initial loading phase due to muscle paresis [11, 31]. This 4. Discussion results in the tibia not rolling forward over the calcanium, which is observed as non-linearity, to complete the transfer Moderate to severe ABIs may result in gait and balance defi- cits such as reduced speed, step length, and abnormal loading of the body weight from the contralateral limb [11]. In this and unloading characteristics during walking. Current study, linearity of loading was quantified using goodness of fit, which showed that the individuals with ABI had a statis- research is focused on reducing these deficits with the use of REs and understanding the therapeutic effects of REs in tically significant improvement in the linearity of their load- the chronic stages of recovery. Recovery may be gradual dur- ing responses at follow-up compared to baseline (Figure 2- ing the chronic stages and may take place over the course of P2, Figure 3-P4, and Table 3). This may indicate that the par- several years. In this study, the efficacy of RE usage for 12 ses- ticipants have healthier foot strike, and initial loading response with improved momentum, at follow-up compared sions on the recovery of gait in adolescents and young adults with chronic ABI was investigated using kinetic and to baseline. The slope of the initial loading response showed temporal-spatial outcomes, such as loading, unloading, an increase at follow-up compared to baseline (Table 3). speed, step length, and gait cycle timing. Following the use Although the difference was not significant, it had a high of the RE, an increase in linearity of loading during IDS with Cohen’s d effect size at follow-up compared to baseline, indicating that the participants had improved their rate of an associated increase in speed, step length, and decrease in stance phase time was observed. loading, once again, demonstrating preserved forward During a healthy gait cycle, the loading/unloading momentum. Thus, at follow-up, TVP showed an increase in response is bilaterally symmetrical, as is observed in smooth linearity and rate of loading similar to healthy indi- Figure 2. It has three distinct phases. The role of the first viduals [17]. Research has shown that improved linearity and slope during loading result in improved gait pattern, phase is the weight acceptance which includes initial contact (heel strike) and the initial loading response. This is the ini- resulting in improved functional ambulation and quality of tial double support phase of the gait cycle and starts with heel life [14]. Also, improvements in initial loading may reduce strike and continues until the contralateral foot is in swing breaking forces [32], leading to a more efficient gait. Though [17]. The second phase supports the upper body on a single the participant’s overall loading profiles showed improve- ments from baseline to follow-up, they still preferentially limb, and it is comprised of midstance and terminal stance (single support phase). The third phase includes preswing loaded their unaffected side. and the terminal unloading phase [17]. Distinct peaks and a The loading profile with the RE showed a smooth linear valley are observed in the loading profile during these phases, loading (Figure 3), but without any distinctive peaks. A pas- with a peak force (Figure 2-VP1) at the end of the initial load- sive ankle in the RE (which provides a midfoot or flat foot landing) or lower speed may have resulted in the absence of ing, followed by a midstance and terminal phase where there is a decrease in loading (Figure 2-VP2) and ending with two distinctive peaks. Changes in walking speed may another peak force (Figure 2-VP3) before terminal unloading improve this profile and need to be investigated further. In [29]. In addition, bilaterally symmetrical loading profiles are addition, a bilaterally symmetrical profile with lower variabil- present. These attributes are observed in the reference HC ity was observed at both baseline and follow-up while walk- ing with the RE compared to walking without the RE, gait. In contrast, the individuals with ABI had a perturbation 8 Applied Bionics and Biomechanics Step length-unaffected Step length-affected 0 0 12 3 4 5 6 7 12 3 4 5 6 7 Subjects Subjects Baseline - without RE Follow-up - with RE Baseline - without RE Follow-up - with RE Follow-up without RE Follow-up without RE HC HC Baseline - with RE Baseline - with RE (a) (b) Speed 1.4 1.2 0.8 0.6 0.4 0.2 0.0 12 3 4 5 6 7 Subjects Baseline - without RE Follow-up - with RE Follow-up without RE HC Baseline - with RE (c) Figure 4: Mean ± standard error of the step length with and without the RE on the (a) affected side, (b) unaffected side, and (c) walking speed. indicating that the RE provides a consistent training environ- ambulation speed (1 m/s) [35]. An increase in speed could improve their community ambulation and by extension, ment throughout therapy. Thus, the improvements in the smoothness and linearity of loading could be attributed to their activities of daily life. Previous research on overground the RE training. Our results are in accordance with previous gait training with RE has shown the ability of RE to improve research which showed that with the use of overground RE, balance and functional ambulation in patients with ABI [36, there was an increased bilateral limb loading symmetry that 37]. Overground REs require active participation, where closely resembled able-bodied gait [33, 34]. patients initiate each step and are responsible for maintain- In addition, after 12 sessions of RE gait training, there ing trunk and balance [21, 37]. Active participation in com- was a slight increase in the step length and an increase in bination with RE’s ability to provide quality gait and walking speed (Figure 4(a)). There was no change in swing increased dose training promotes improved brain plasticity time and a decrease in stance time. These results suggest that and connectivity remodulation, as compared to conventional the increased step length (Figure 4(a)) and improved TVP gait training [21, 38, 39]. profile may have contributed to the increased walking speed This is one of the first studies to show the feasibility of (Figure 2(a)). They are significant since most of the partici- using RE for gait training in adolescents and young adults pants in this study were walking below the community with ABI. The results from this preliminary study show a Length (cm) Speed (m/s) Applied Bionics and Biomechanics 9 [6] N. N. Byl, E. A. Pitsch, and G. M. Abrams, “Functional out- therapeutic effect of RE on the loading/unloading character- comes can vary by dose: learning-based sensorimotor training istics and a consequent impact on functional ambulation. for patients stable poststroke,” Neurorehabilitation and Neural Although these results are promising, the limitations of this Repair, vol. 22, no. 5, pp. 494–504, 2008. investigation are the limited sample size, number of training [7] J. D. Schaechter, “Motor rehabilitation and brain plasticity sessions, and absence of control group. The data from this after hemiparetic stroke,” Progress in Neurobiology, vol. 73, study indicates some promising results for therapeutic effects no. 1, pp. 61–72, 2004. of an RE device for ABI gait rehabilitation that should con- [8] M. Pekna, M. Pekny, and M. Nilsson, “Modulation of neural tinue to be explored with a larger sample. plasticity as a basis for stroke rehabilitation,” Stroke, vol. 43, no. 10, pp. 2819–2828, 2012. 5. Conclusion [9] S. Lennon, D. Baxter, and A. Ashburn, “Physiotherapy Based on the Bobath Concept in Stroke Rehabilitation: A Survey The results from this investigation suggest that improvement within the UK,” Disability and Rehabilitation, vol. 23, no. 6, in functional and neuromechanical outcomes after 4 weeks of pp. 254–262, 2001. RE gait training can be achieved in adolescents and young [10] D. R. Louie and J. J. Eng, “Powered robotic exoskeletons in adults with chronic ABI. This study suggests that there could post-stroke rehabilitation of gait: a scoping review,” Journal be potential long-term effects of improved loading and of Neuro Engineering and Rehabilitation, vol. 13, no. 1, 2016. unloading, increased step length, and increased speed due [11] D. A. Winter, Biomechanics and Motor Control of Human to RE gait training. While the current results are promising, Gait: Normal, Elderly, and Pathological, Waterloo Biomechan- future studies with a larger sample would be required to fur- ics, Canada, 2009. ther understand the efficacy of the RE in adolescents and [12] M. T. Jahnke, S. Hesse, C. Schreiner, and K. H. Mauritz, young adults to confirm any training effect conclusively. “Dependences of Ground Reaction Force Parameters on Habitual Walking Speed in Hemiparetic Subjects,” Gait Pos- Data Availability ture, vol. 3, no. 1, pp. 3–12, 1995. [13] K. Karunakaran, N. Ehrenberg, J. Cheng, and K. J. Nolan, The data is currently unavailable due to Kessler Foundation “Effects of Robotic Exoskeleton Gait Training on an Adoles- IRB restrictions. cent with Brain Injury,” in IEEE 41h Annual Interna- tional Conference on Engineering in Medicine and Biology Society (EMBC), pp. 4445–4448, Berlin, Germany, Conflicts of Interest Germany, 2019. The authors have no conflict of interest. [14] K. J. Nolan and M. Yarossi, “Weight Transfer Analysis in Adults with Hemiplegia Using Ankle Foot Orthosis,” Pros- thetics and orthotics international, vol. 35, no. 1, pp. 45– Acknowledgments 53, 2017. We would like to acknowledge Danielle Nisenson, Kathleen [15] C. Chen, P. W. Hong, C. Chen et al., “Ground Reaction Force Chervin, and Brandon Ross for their assistance during this Patterns in Stroke Patients with Various Degrees of Motor Recovery Determined by Plantar Dynamic Analysis,” Chang study. Research supported by the New Jersey Health Founda- Gung medical journal, vol. 30, no. 1, 2007. tion PC5-18, Children’s Specialized Hospital, the Kessler [16] A. M. Wong, Y. C. Pei, W. H. Hong, C. Y. Chung, Y. C. Lau, Foundation, and the Reitman Foundation. and C. P. Chen, “Foot Contact Pattern Analysis in Hemiplegic Stroke Patients: An Implication for Neurologic Status Deter- References mination,” Archives of physical medicine and rehabilitation, vol. 85, no. 10, pp. 1625–1630, 2004. [1] R. A. Newton, “Balance abilities in individuals with moderate [17] J. Perry and J. Burnfield, “GAIT Normal and Pathological and severe traumatic brain injury,” Brain Injury, vol. 9, no. 5, Function,” Journal of Sports Science and Medicine, vol. 9, pp. 445–451, 2009. no. 2, p. 353, 2010. [2] G. Williams, M. E. Morris, A. Schache, and P. R. McCrory, [18] H. J. Dananberg, “Functional Hallux Limitus and its Relation- “Incidence of Gait Abnormalities after Traumatic Brain ship to Gait Efficiency,” Journal of the American Podiatric Injury,” Archives of physical medicine and rehabilitation, Medical Association, vol. 76, no. 11, pp. 648–652, 1986. vol. 90, no. 4, pp. 587–593, 2009. [19] P. A. Goldie, T. A. Matyas, O. M. Evans, M. Galea, and T. M. [3] K. M. Michael, J. K. Allen, and R. F. MacKo, “Reduced Ambu- Bach, “Maximum Voluntary Weight-Bearing by the Affected latory Activity after Stroke: The Role of Balance, Gait, and Car- and Unaffected Legs in Standing Following Stroke,” Clinical diovascular Fitness,” Archives of physical medicine and Biomechanics, vol. 11, no. 6, pp. 333–342, 1996. rehabilitation, vol. 86, no. 8, pp. 1552–1556, 2005. [20] A. Esquenazi, M. Talaty, and A. Jayaraman, “Powered exoskel- [4] M. A. Kemu, “Kinetics and kinematics of loading response in stroke patients (a review article),” Annals of King Edward Med- etons for walking assistance in persons with central nervous system injuries: a narrative review,” PM&R, vol. 9, no. 1, ical University, vol. 14, no. 4, 2008. pp. 46–62, 2017. [5] M. G. Benedetti, V. Agostini, M. Knaflitz, V. Gasparroni, M. Boschi, and R. Piperno, “Self-Reported Gait Unsteadiness [21] F. Molteni, G. Gasperini, G. Cannaviello, and E. Guanziroli, in Mildly Impaired Neurological Patients: An Objective “Exoskeleton and end-effector robots for upper and lower Assessment through Statistical Gait Analysis,” Journal of neu- limbs rehabilitation: narrative review,” PM&R, vol. 10, 9 Suppl roengineering and rehabilitation, vol. 9, no. 1, p. 64, 2012. 2, pp. S174–S188, 2018. 10 Applied Bionics and Biomechanics [38] R. S. Calabrò, A. Naro, M. Russo et al., “Shaping Neuroplasti- [22] H. Igo Krebs, N. Hogan, M. L. Aisen, and B. T. Volpe, “Robot- aided neurorehabilitation,” IEEE Transactions on Rehabilita- city by Using Powered Exoskeletons in Patients with Stroke: A tion Engineering, vol. 6, no. 1, pp. 75–87, 1998. Randomized Clinical Trial,” Journal of neuroengineering and rehabilitation, vol. 15, no. 1, p. 35, 2018. [23] A. M. Dollar and H. Herr, “Lower extremity exoskeletons and active orthoses: challenges and state-of-the-art,” IEEE Trans- [39] G. J. Androwis, R. Pilkar, A. Ramanujam, and K. J. Nolan, “Electromyography Assessment during Gait in a Robotic Exo- actions on Robotics, vol. 24, no. 1, pp. 144–158, 2008. skeleton for Acute Stroke,” Frontiers in neurology, vol. 9, 2018. [24] K. Anam and A. A. Al-Jumaily, “Active exoskeleton control systems: state of the art,” Procedia Engineering, vol. 41, pp. 988–994, 2012. [25] L. Dipietro, M. Ferraro, J. J. Palazzolo, H. I. Krebs, B. T. Volpe, and N. Hogan, “Customized interactive robotic treatment for stroke: EMG-triggered therapy,” IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. 13, no. 3, pp. 325– 334, 2005. [26] A. Jayaraman, S. Burt, and W. Z. Rymer, “Use of lower-limb robotics to enhance practice and participation in individuals with neurological conditions,” in Pediatric Physical Therapy, vol. 29, pp. S48–S56, 2017. [27] D. J. Reinkensmeyer, J. L. Emken, and S. C. Cramer, “Robotics, Motor Learning, and Neurologic Recovery,” Annual review of biomedical engineering, vol. 6, no. 1, pp. 497–525, 2004. [28] US F& DAdministration, Ekso User Manual, 2020, https:// www.accessdata.fda.gov/cdrh_docs/pdf20/K200574.pdf. [29] S. Winiarski and A. Rutkowska-Kucharska, “Estimated Ground Reaction Force in Normal and Pathological Gait,” Acta of Bioengineering & Biomechanics, vol. 11, no. 1, [30] D. A. Winter, Biomechanics and Motor Control of Human Movement, John Wiley & Sons., Hoboken, New Jersey, 2009. [31] K. Fujita, H. Hori, and Y. Kobayashi, “Contribution of Muscle Activity at Different Gait Phases for Improving Walking Per- formance in Chronic Stroke Patients with Hemiparesis,” Jour- nal of physical therapy science, vol. 30, no. 11, pp. 1381–1385, [32] T. S. Keller, A. M. Weisberger, J. L. Ray, S. S. Hasan, R. G. Shiavi, and D. M. Spengler, “Relationship between Vertical Ground Reaction Force and Speed during Walking, Slow Jog- ging, and Running,” Clinical biomechanics, vol. 11, no. 5, pp. 253–259, 1996. [33] S. R. Husain, A. Ramanujam, K. Momeni, and G. F. Forrest, “Effects of exoskeleton training intervention on net loading force in chronic spinal cord injury,” in Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, Honolulu, HI, USA, 2018. [34] A. Ramanujam, A. Spungen, P. Asselin et al., “Training response to longitudinal powered exoskeleton training for SCI,” in Biosystems and Biorobotics, 2017. [35] N. M. Salbach, K. O'Brien, D. Brooks et al., “Speed and dis- tance requirements for community ambulation: a systematic review,” Archives of Physical Medicine and Rehabilitation, vol. 95, no. 1, pp. 117–128.e11, 2014. [36] A. Rojek, A. Mika, Ł. Oleksy, A. Stolarczyk, and R. Kielnar, “Effects of Exoskeleton Gait Training on Balance, Load Distri- bution, and Functional Status in Stroke: A Randomized Con- trolled Trial,” Frontiers in Neurology, vol. 10, 2020. [37] F. Molteni, G. Gasperini, M. Gaffuri et al., “Wearable Robotic Exoskeleton for Overground Gait Training in Sub-Acute and Chronic Hemiparetic Stroke Patients: Preliminary Results,” European journal of physical and rehabilitation medicine, vol. 53, 2017. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Applied Bionics and Biomechanics Hindawi Publishing Corporation

Kinetic Gait Changes after Robotic Exoskeleton Training in Adolescents and Young Adults with Acquired Brain Injury

Loading next page...
 
/lp/hindawi-publishing-corporation/kinetic-gait-changes-after-robotic-exoskeleton-training-in-adolescents-5Mkg3ozC5N
Publisher
Hindawi Publishing Corporation
Copyright
Copyright © 2020 Kiran K. Karunakaran et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
ISSN
1176-2322
eISSN
1754-2103
DOI
10.1155/2020/8845772
Publisher site
See Article on Publisher Site

Abstract

Hindawi Applied Bionics and Biomechanics Volume 2020, Article ID 8845772, 10 pages https://doi.org/10.1155/2020/8845772 Research Article Kinetic Gait Changes after Robotic Exoskeleton Training in Adolescents and Young Adults with Acquired Brain Injury 1,2,3 1,3 2,3 Kiran K. Karunakaran , Naphtaly Ehrenberg , JenFu Cheng , 2,3 1,2,3 Katherine Bentley, and Karen J. Nolan Center for Mobility and Rehabilitation Engineering Research, Kessler Foundation, West Orange, New Jersey 07052, USA Physical Medicine and Rehabilitation, Rutgers New Jersey Medical School, Newark, New Jersey 07103, USA Research Department, Children's Specialized Hospital, New Brunswick, New Jersey 08901, USA Correspondence should be addressed to Kiran K. Karunakaran; kkarunakaran@kesslerfoundation.org Received 15 July 2020; Revised 11 September 2020; Accepted 8 October 2020; Published 28 October 2020 Academic Editor: Dongming Gan Copyright © 2020 Kiran K. Karunakaran et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Background. Acquired brain injury (ABI) is one of the leading causes of motor deficits in children and adults and often results in motor control and balance impairments. Motor deficits include abnormal loading and unloading, increased double support time, decreased walking speed, control, and coordination. These deficits lead to diminished functional ambulation and reduced quality of life. Robotic exoskeletons (RE) for motor rehabilitation can provide the user with consistent, symmetrical, goal-directed repetition of movement, as well as balance and stability. Purpose. The goal of this preliminary prospective before and after study is to evaluate the therapeutic effect of RE training on the loading/unloading and spatial-temporal characteristics in adolescents and young adults with chronic ABI. Method. Seven participants diagnosed with ABI between the ages of 14 and 27 years participated in the study. All participants received twelve 45 minute sessions of RE gait training. The bilateral loading (linearity of loading and rate of loading), speed, step length, swing time, stance time, and total time were collected using Zeno™ walkway (ProtoKinetics, Havertown, PA, USA) before and after RE training. Results. Results from the study showed improved step length, speed, and an overall progression towards healthy bilateral loading, with linearity of loading showing a significant therapeutic effect (p <0:05). Conclusion. These preliminary results suggest that high dose, repetitive, consistent gait training using RE has the potential to induce recovery of function in adolescents and young adults diagnosed with ABI. 1. Introduction needed for a functional recovery [10]. Consequently, after motor rehabilitation individuals with ABI can experience Acquired brain injury (ABI) is a leading cause of hemiparesis variable recovery with residual gait deviations. As a result resulting in gait and balance deficits in adolescents and young of these residual motor and balance deficits, individuals diag- adults [1–3]. These deficits result in abnormal loading and nosed with ABI may develop compensatory mechanisms unloading, increased double support time, and decreased such as abnormal or asymmetrical loading and unloading walking speed, control, and coordination, leading to impair- characteristics and prolonged weight transfer, in order to ment in functional ambulation and associated activities of achieve ambulation [4, 11]. daily living [2–5]. Recovery involves rehabilitation through Ground reaction force (GRF) parameters, e.g., peak high dose physical therapy [6]. It is based on the theory that values, linearity, and rates of loading/unloading, symmetry the human brain is capable of self-reorganization, or plastic- coefficients, and force integrals, are used for assessment of pathological gait and evaluation of therapeutic efficacy ity, through continuous, consistent, repeated practice aimed at restoring function and independence [7–9]. Conventional [12–14]. The analysis of vertical forces during gait provides therapy alone may not be able to provide enough consistent information on weight bearing and balance, and these force mass practice and repetition to facilitate the neuroplasticity or pressure patterns in the paretic leg are known to be 2 Applied Bionics and Biomechanics Table 1: Subject demographics. Subject Condition Affected side Gender Age Height (m) Weight (Kg) Years since injury 1 TBI Left Male 23 1.83 90.00 1 2 Stroke Left Female 23 1.60 54.00 .9 3 Anoxic Left Female 17 1.63 54.00 2 4 Stroke Left Male 16 1.83 67.50 .5 5 TBI Left Male 22 1.78 63.50 3 6 TBI Left Female 27 1.63 62.60 5 7 TBI Right Male 14 1.65 49.44 0.5 Mean ± SD 20:29 ± 4:68 1:71 ± 0:10 63:01 ± 13:51 1:84 ± 1:66 1 HC n/a Male 26 1.57 77.11 n/a of the RE gait training environment for reference, to help correlated with walking speed and motor recovery in stroke patients [15, 16]. Therefore, improvements in vertical pres- understand the therapeutic effects. sure profiles might demonstrate improvement in gait and overall functional ambulation. 2. Materials and Methods In a healthy gait cycle, the loading response is bilaterally symmetrical [17]. It has three distinct phases, starting with This is a preliminary prospective before and after study a peak force during the initial loading phase, followed by a which is to evaluate the therapeutic effect of RE training on midstance phase where there is a decrease in loading, and the loading/unloading and spatial-temporal characteristics ending with another peak force during the terminal unload- in adolescents and young adults with chronic ABI. ing phase [17]. Thus, the loading response resembles a bimodal M shape [17]. The transfer of weight during the 2.1. Participants. Twelve participants with ABI were initial loading phase allows for the efficient transfer of recruited for this study. Kinetic and temporal-spatial data momentum from one leg to the other and the uninter- for baseline and follow-up visits were only available from rupted use of the kinetic energy that is created by swing nine participants. Further, two participants were excluded limb activity [18]. The rate of loading determines the speed from the analysis as they were diagnosed with bilateral defi- of gait [15]. Research has shown that smooth linear loading cits. Therefore, seven participants diagnosed with ABI and helps conserve momentum and shifts the body’s weight to hemiparesis and a single healthy control (HC) were included the next phase of the gait cycle while maintaining speed for this analysis (Table 1). The inclusion criteria for ABI par- [14, 17, 19]. If weight transfer is delayed or loading is non- ticipants in this investigation included (1) diagnosed with an linear, it will result in a loss of energy and hence less effi- acquired brain injury (anoxic, stroke, or TBI); (2) between cient gait, leading to decreased speed and necessitating the the ages of 13 and 28; (3) able to walk with or without an use of other compensatory gait mechanisms. Limb loading assistive device; (4) no additional orthopedic, neuromuscu- is often inefficient in individuals with ABI because the lar, or severe neurological pathologies (unrelated to their affected limb has a difficult time accepting weight, resulting ABI) that would interfere with their ability to walk; (5) able in a nonlinear loading and decreased momentum. Conse- to stand upright for 30 minutes with assistance; and (6) able quently, a targeted goal during gait rehabilitation in ABI to physically fit into the RE (weight ≤ 100 kg, height ≤ 1.88 m, is to improve loading profiles in order to facilitate healthier hip width 0.36 m-0.46 m). The inclusion criteria for the HC movement patterns. were no orthopedic, neuromuscular, or severe neurological Lower extremity robotic exoskeletons (REs) are currently pathologies that would interfere with gait and balance. All being used in rehabilitation to restore gait functionality. REs procedures performed in this investigation were approved have the potential to provide the user with high dose, consis- by the Human Subjects Institutional Review Board at Kessler tent, symmetrical bilateral loading [9, 20–26] profiles during Foundation, and informed consent was obtained prior to gait. It can provide an increased number of steps (increased study participation. dosing) in a consistent and controlled training environment, which is ideal for inducing neuroplasticity [20, 27]. This is 2.2. Robotic Exoskeleton Gait Training. Robotic gait training essential for chronic ABI patients who need high dose, con- in bilateral assistance mode was administered as an outpa- sistent therapy to induce cortical reorganization for func- tient rehabilitation gait intervention at the Kessler Founda- tional recovery. This study utilized an RE to provide tion using a commercially available robotic exoskeleton intensive gait training to adolescents and young adults diag- (Figure 1, EksoGT, Ekso Bionics, Inc., Richmond, CA, nosed with an ABI, with the goal of evaluating the efficacy of USA) for 45 minutes per day for 12 days over a period of 4 high repetition robotic training on loading/unloading pro- weeks. A licensed physical therapist administered all RE gait files. The objective of this preliminary prospective study training sessions and adjusted the assistance provided by the was to evaluate the therapeutic effect of RE on their loadin- robot to the individual participant’s therapy progression. A g/unloading characteristics. This paper also provides details member of the study team was present at all times during Applied Bionics and Biomechanics 3 Figure 1: Robotic gait training with a participant with ABI administered by a trained physical therapist. lateral weight shift to a predetermined distance from mid- the RE gait training sessions to assist the physical therapist and ensure participant safety. The healthy control participant line to initiate advancement of the back (preswing) leg. was not given RE training as part of the investigation. The RE provided overground gait rehabilitation under 2.3. Data Collection Procedures. The participants with ABI the guidance of a licensed physical therapist. The RE’s participated in two data collection sessions (baseline and upper section is attached to the user’s upper body, with follow-up after 12 sessions of outpatient RE training), and a backpack style shoulder harness and torso brace and also data was collected while walking with (training environment) houses the battery pack, while the lower sections are and without the RE. The participants with ABI did not utilize affixed to the legs with upper thigh straps, shin guards, any orthotic devices or dorsiflexion wrap during data collec- and secure foot bindings. The RE has two active degrees tion. The reference HC participated in one data collection of freedom at the hip and the knee, respectively, and a session, and data was collected while walking without the RE. passively sprung ankle joint with adjustable stiffness in During each data collection session, temporal, spatial, the sagittal plane. The RE provides assistive torques to and loading data was collected using a Zeno™ walkway (Pro- the hip and knee joints to perform the predefined gait tra- toKinetics, Havertown, PA, USA) at 120 Hz. The participant jectory and provides variable assistance as required by the performed up to 6 walking trials (approximately six 10 meter participants bilaterally. The actuated range of motion at walks) at a self-selected pace per condition (with and without ° ° the hip is -20 to 135 , and the actuated range for the knee RE) and wore shoes for all walking assessments. Participants ° ° is 0 to 120 [28]. All steps were initiated by the partici- were allowed to rest or take breaks at any time during testing pant. In order to trigger a step, the subject performed a to minimize the effects of fatigue. A member of the study 4 Applied Bionics and Biomechanics Table 2: Outcome measures. Outcome measure Description Statistical analysis TVP during the stance phase of each gait cycle was computed, and an average TVP for all gait cycles was calculated for both the legs without and with RE that was computed for each subject and for the HC. The TVP was normalized to 100% of stance in each condition for comparison. The stance phase was further divided into Total vertical initial double support (IDS) and terminal double support pressure (TVP) (TDS) phases based on heel strike and toe off. The IDS pressure was computed as the pressure between ipsilateral heel strikes to contralateral toe off. The TDS pressure was computed as the pressure between contralateral heel strikes to ipsilateral toe off. Mean and standard deviation TVP for all participants with ABI was computed. A best fit line was computed for the average IDS loading phase for each session for each subject. A goodness of the fit was computed to assess the error between the fitted line and the average loading during IDS for each subject in Kolmogorov- Smirnov Z test (p <0:05) of normality Linearity of each session. showed that the data were not normal. Wilcoxson signed loading (goodness The goodness of fit was used to assess the smooth linearity rank test was used to determine the therapeutic effect of fit) of loading. R-square (R ) was computed to assess the (baseline to follow-up without RE) on goodness of fit. square of the correlation between average loading during IDS and the best fit line. A higher R value signifies a closer fit to the best fit line or increased linearity. Kolmogorov- Smirnov Z test (p >0:05) of normality The slope of the average IDS loading phase was computed Rate of linear showed that the data were normal. A paired sample for each subject. Slope indicates the rate of linear loading. loading (slope of t-test was performed to determine the therapeutic effect Increased slope in the IDS phase indicates an increased initial loading) (baseline to follow-up without RE) on the slope of initial moment during the first rocker. loading. Kolmogorov- Smirnov Z test (p > :05) of normality The average walking speed was computed for each subject showed that the data were normal. A paired sample Walking speed as the linear distance with respect to time to complete a t-test was performed to determine the therapeutic effect gait cycle. (baseline to follow-up without RE) on walking speed. The average step length for each gait cycle was computed Kolmogorov- Smirnov Z test (p > :05) of normality as the forward linear displacement between foot contact of showed that the data were normal. A paired sample Step length the ipsilateral leg to foot contact of the contralateral leg t-test was performed to determine the therapeutic effect during each gait cycle. Average step length was computed (baseline to follow-up without RE) on step length. for each subject. Total time was computed as the time between foot contact of one leg to the subsequent foot contact of the same leg. The average total time was computed for each gait cycle. Kolmogorov- Smirnov Z test (p < :05) of normality Further, average swing time for each subject during each showed that the data were not normal. Wilcoxson signed Temporal condition was computed as the time between the foot off rank test was used to the therapeutic effect (baseline to measures the floor of one leg to foot contact of the same leg during follow-up without RE) on total time, swing time, and the gait cycle. Average stance time for each subject during stance time. each condition was computed as the time between the foot contact of one leg to toe off the same leg during the gait cycle. team was present with the participants at all times during the ized pressure data were exported. Custom MATLAB algo- rithms were used to analyze the gait trajectories for each walking trials. During the baseline and follow-up gait assessments in the session with and without the RE. The exported data were RE, torque was provided bilaterally as needed at the hip and further divided into gait cycles, with a gait cycle being knee to complete the gait. defined as the period from ground contact of one foot to the subsequent ground contact of the same foot. Data for up to 15 gait cycles per condition and per session were 2.4. Data Analysis. PKMAS (ProtoKinetics Havertown, PA, USA) and MATLAB (MathWorks, Natick, MA, USA) were available for each subject and were used for data analysis. used for data analysis. The data were preprocessed using Selected outcome measures and statistical analyses are pre- the PKMAS software, and temporal, spatial, and normal- sented in Table 2. Applied Bionics and Biomechanics 5 Baseline Follow-up Healthy control Without RE Without RE Without RE 400 400 400 VP1 VP3 300 300 300 VP2 P2 P1 200 200 200 100 100 100 0 0 0 0 50 100 0 50 100 0 50 100 With RE With RE Stance phase in percentage Right Left 500 500 400 400 300 300 200 200 100 100 0 0 0 50 100 0 50 100 Stance phase in percentage Affected Unaffected Figure 2: Mean ± standard deviation of the TVP of the affected and unaffected leg of individuals with ABI during walking with and without an RE at baseline and follow-up and one reference HC. Data is normalized to 100% of the stance phase. 3. Results deviation from the mean across gait cycles for participants with ABI was smaller in the RE training environment com- 3.1. Total Vertical Pressure (TVP). Figure 2 shows the average pared to without the RE at both baseline and follow-up. TVP of all participants with ABI at baseline and at follow-up 3.2. Rate of Linear Loading (Slope of Initial Loading). There for both the affected and the unaffected sides with and was an increase in the average slope from baseline to without RE and one reference HC. The reference HC’s TVP follow-up while walking without the RE (Table 3). A paired profile demonstrated linear loading and unloading, a sym- sample t-test for average slope did not show a significant metrical loading pattern during stance bilaterally, and a min- therapeutic effect between baseline and follow-up, though imal deviation from the mean across gait cycles. For the effect size (p =0:136, Cohen’s d effect size was 0.59) was participants with ABI walking without RE at baseline, the high (Table 3). The average slope was higher in the training TVP demonstrated a perturbation during loading for the environment (with the RE) both at baseline and at follow- IDS phase (Figure 2-P1and Figure 3-P3) and increased load- up compared to without the RE. The RE training environ- ing on the unaffected side. At follow-up, the perturbation ment was similar to HC (Table 3). during loading for IDS decreased (Figure 2-P2 and Figure 3-P3), and there was increased loading on the affected 3.3. Linearity of Loading (Goodness of Fit). There was an limb during midstance. There was increased variation in the increase in average R from baseline to follow-up while walk- TVP on the affected side compared to the unaffected side for ing without the RE (Table 3). The Wilcoxson signed rank test individuals with ABI, demonstrated by the standard deviations. In the RE training environment, the TVP profile showed for R showed a significant therapeutic effect between base- line and follow-up, and the effect size (p =0:018, Cohen’s linear loading during IDS, but did not show the distinctive peaks as observed in the HC’s data (Figure 2-VP1 and d effect size was 0.6334) was high. The R of loading Figure 2-VP3) in both the affected and unaffected sides at was higher in the training environment (with the RE) at both baseline and follow-up. Furthermore, a bilaterally sym- baseline and about the same at follow-up compared to metrical loading profile was observed in the RE training envi- without the RE. The RE training environment was similar ronment and for the HC as compared to without RE. The to the HC (Table 3). Vertical pressure 6 Applied Bionics and Biomechanics Baseline Follow-up Healthy control Without RE Without RE Without RE 400 400 300 300 300 P3 P4 200 200 200 100 100 0 0 0 50 100 0 50 100 0 50 100 With RE With RE IDS phase in percentage Right Left 500 500 Best fit-right Best fit-left 400 400 300 300 200 200 100 100 0 0 0 50 100 0 50 100 IDS phase in percentage Affected Unaffected Best fit-affected Best fit-unaffected Figure 3: Mean ± standard deviation of the initial loading phase of the right and left leg of subjects with ABI while walking with and without the RE at baseline and at follow-up and a reference HC. The dotted lines are the best fit lines for the loading profile. Table 3: Mean ± standard error of initial double support (IDS) loading characteristics on the affected side of all participants with ABI and IDS loading characteristics on the left side of one HC. Metric Baseline-without RE Follow-up-without RE Baseline-with RE Follow-up-with RE HC 2:33 ± 0:28 2:60 ± 0:36 3:44 ± 0:29 3:43 ± 0:24 Slope 3.46 0:889 ± 0:05 0:934 ± 0:04 0:927 ± 0:02 0:923 ± 0:02 Goodness of fit 0.99 3.4. Step Length. Step length without the RE showed an change (Figure 2(c)). A paired sample t-test did not show a increase in both the unaffected and affected sides at follow- significant therapeutic effect between baseline and follow- up, though the effect size (p =0:083, Cohen’s d effect size up compared to baseline in subjects 2,3,5,6, and 7, while it showed a decrease in subject 1 and no change on the unaf- was 0.80) was high. The average walking speed was lower in fected side and an increase on the affected side in subject 4 the training environment (with the RE) compared to without (Figures 2(a) and 2(b)). A paired sample t-test did not show the RE at baseline and follow-up. The average speed of the a significant therapeutic effect between baseline and follow- HC data was higher than ABI without RE. up, though the effect size (p = :075, Cohen’s d effect size 3.6. Temporal Characteristics. Average total time, stance was 0.81) was high. The average step length was lower in time, and total double support time decreased from baseline the training environment (with the RE) compared to without to follow-up (Table 4), but did not show a significant the RE at baseline and follow-up. The average step length of therapeutic effect (p =0:612, p =0:398, the HC data was higher than ABI without RE. totaltime stance time p =0:237). The average total time, stance total double support time 3.5. Walking Speed. Walking speed without the RE increased time, and total double support time were higher in the train- from baseline to follow-up for subjects 2,4,5,6, and 7, while ing environment (with the RE) compared to without the RE subject 1 decreased their speed, and subject 3 showed no at baseline and at follow-up. Average total time, stance time, Vertical pressure Applied Bionics and Biomechanics 7 Table 4: Mean ± standard error of temporal characteristics. Metric Baseline-without RE Follow-up-without RE Baseline-with RE Follow-up-with RE HC Total time 2:18 ± 0:69 1:98 ± 0:50 3:78 ± 0:17 3:35 ± 0:09 1.24 1:66 ± 0:62 1:46 ± 0:45 3:12 ± 0:15 2:69 ± 0:10 Stance time 0.80 Affected 0:52 ± 0:07 0:52 ± 0:06 0:66 ± 0:05 0:66 ± 0:06 Swing time 0.44 1:25 ± 0:62 1:05 ± :045 2:40 ± 0:17 2:04 ± 0:10 Total double support 0.36 2:15 ± 0:66 1:95 ± 0:48 3:77 ± 0:16 3:33 ± 0:10 Total time 1.27 1:74 ± 0:66 1:55 ± 0:47 3:05 ± 0:17 2:69 ± 0:10 Stance time 0.85 Unaffected 0:41 ± 0:04 0:40 ± 0:04 0:72 ± 0:11 0:64 ± 0:02 Swing time 0.43 0:73 ± 0:49 0:44 ± 0:21 1:19 ± 0:10 1:01 ± 0:05 Total double support 0.23 during initial loading (Figure 2-P1, Figure 3-P3) on the and total double support time of the HC data were lower than ABI without RE. affected side during baseline, which indicates a nonlinear ini- Swing time did not change (p =0:866) from baseline to tial loading response. Linear loading directly contributes to follow-up (Table 4). The average swing time was higher in the momentum during gait [30]. Therefore, any nonlinearity the training environment (with the RE) compared to without would result in decreased momentum and slower load trans- the RE at baseline and follow-up. Average swing time of the fer during gait, leading to a disruption in forward progres- HC data was lower than ABI without RE. sion. ABI patients with hemiplegia often present with reduced hip flexion, dorsiflexion, and plantarflexion during the initial loading phase due to muscle paresis [11, 31]. This 4. Discussion results in the tibia not rolling forward over the calcanium, which is observed as non-linearity, to complete the transfer Moderate to severe ABIs may result in gait and balance defi- cits such as reduced speed, step length, and abnormal loading of the body weight from the contralateral limb [11]. In this and unloading characteristics during walking. Current study, linearity of loading was quantified using goodness of fit, which showed that the individuals with ABI had a statis- research is focused on reducing these deficits with the use of REs and understanding the therapeutic effects of REs in tically significant improvement in the linearity of their load- the chronic stages of recovery. Recovery may be gradual dur- ing responses at follow-up compared to baseline (Figure 2- ing the chronic stages and may take place over the course of P2, Figure 3-P4, and Table 3). This may indicate that the par- several years. In this study, the efficacy of RE usage for 12 ses- ticipants have healthier foot strike, and initial loading response with improved momentum, at follow-up compared sions on the recovery of gait in adolescents and young adults with chronic ABI was investigated using kinetic and to baseline. The slope of the initial loading response showed temporal-spatial outcomes, such as loading, unloading, an increase at follow-up compared to baseline (Table 3). speed, step length, and gait cycle timing. Following the use Although the difference was not significant, it had a high of the RE, an increase in linearity of loading during IDS with Cohen’s d effect size at follow-up compared to baseline, indicating that the participants had improved their rate of an associated increase in speed, step length, and decrease in stance phase time was observed. loading, once again, demonstrating preserved forward During a healthy gait cycle, the loading/unloading momentum. Thus, at follow-up, TVP showed an increase in response is bilaterally symmetrical, as is observed in smooth linearity and rate of loading similar to healthy indi- Figure 2. It has three distinct phases. The role of the first viduals [17]. Research has shown that improved linearity and slope during loading result in improved gait pattern, phase is the weight acceptance which includes initial contact (heel strike) and the initial loading response. This is the ini- resulting in improved functional ambulation and quality of tial double support phase of the gait cycle and starts with heel life [14]. Also, improvements in initial loading may reduce strike and continues until the contralateral foot is in swing breaking forces [32], leading to a more efficient gait. Though [17]. The second phase supports the upper body on a single the participant’s overall loading profiles showed improve- ments from baseline to follow-up, they still preferentially limb, and it is comprised of midstance and terminal stance (single support phase). The third phase includes preswing loaded their unaffected side. and the terminal unloading phase [17]. Distinct peaks and a The loading profile with the RE showed a smooth linear valley are observed in the loading profile during these phases, loading (Figure 3), but without any distinctive peaks. A pas- with a peak force (Figure 2-VP1) at the end of the initial load- sive ankle in the RE (which provides a midfoot or flat foot landing) or lower speed may have resulted in the absence of ing, followed by a midstance and terminal phase where there is a decrease in loading (Figure 2-VP2) and ending with two distinctive peaks. Changes in walking speed may another peak force (Figure 2-VP3) before terminal unloading improve this profile and need to be investigated further. In [29]. In addition, bilaterally symmetrical loading profiles are addition, a bilaterally symmetrical profile with lower variabil- present. These attributes are observed in the reference HC ity was observed at both baseline and follow-up while walk- ing with the RE compared to walking without the RE, gait. In contrast, the individuals with ABI had a perturbation 8 Applied Bionics and Biomechanics Step length-unaffected Step length-affected 0 0 12 3 4 5 6 7 12 3 4 5 6 7 Subjects Subjects Baseline - without RE Follow-up - with RE Baseline - without RE Follow-up - with RE Follow-up without RE Follow-up without RE HC HC Baseline - with RE Baseline - with RE (a) (b) Speed 1.4 1.2 0.8 0.6 0.4 0.2 0.0 12 3 4 5 6 7 Subjects Baseline - without RE Follow-up - with RE Follow-up without RE HC Baseline - with RE (c) Figure 4: Mean ± standard error of the step length with and without the RE on the (a) affected side, (b) unaffected side, and (c) walking speed. indicating that the RE provides a consistent training environ- ambulation speed (1 m/s) [35]. An increase in speed could improve their community ambulation and by extension, ment throughout therapy. Thus, the improvements in the smoothness and linearity of loading could be attributed to their activities of daily life. Previous research on overground the RE training. Our results are in accordance with previous gait training with RE has shown the ability of RE to improve research which showed that with the use of overground RE, balance and functional ambulation in patients with ABI [36, there was an increased bilateral limb loading symmetry that 37]. Overground REs require active participation, where closely resembled able-bodied gait [33, 34]. patients initiate each step and are responsible for maintain- In addition, after 12 sessions of RE gait training, there ing trunk and balance [21, 37]. Active participation in com- was a slight increase in the step length and an increase in bination with RE’s ability to provide quality gait and walking speed (Figure 4(a)). There was no change in swing increased dose training promotes improved brain plasticity time and a decrease in stance time. These results suggest that and connectivity remodulation, as compared to conventional the increased step length (Figure 4(a)) and improved TVP gait training [21, 38, 39]. profile may have contributed to the increased walking speed This is one of the first studies to show the feasibility of (Figure 2(a)). They are significant since most of the partici- using RE for gait training in adolescents and young adults pants in this study were walking below the community with ABI. The results from this preliminary study show a Length (cm) Speed (m/s) Applied Bionics and Biomechanics 9 [6] N. N. Byl, E. A. Pitsch, and G. M. Abrams, “Functional out- therapeutic effect of RE on the loading/unloading character- comes can vary by dose: learning-based sensorimotor training istics and a consequent impact on functional ambulation. for patients stable poststroke,” Neurorehabilitation and Neural Although these results are promising, the limitations of this Repair, vol. 22, no. 5, pp. 494–504, 2008. investigation are the limited sample size, number of training [7] J. D. Schaechter, “Motor rehabilitation and brain plasticity sessions, and absence of control group. The data from this after hemiparetic stroke,” Progress in Neurobiology, vol. 73, study indicates some promising results for therapeutic effects no. 1, pp. 61–72, 2004. of an RE device for ABI gait rehabilitation that should con- [8] M. Pekna, M. Pekny, and M. Nilsson, “Modulation of neural tinue to be explored with a larger sample. plasticity as a basis for stroke rehabilitation,” Stroke, vol. 43, no. 10, pp. 2819–2828, 2012. 5. Conclusion [9] S. Lennon, D. Baxter, and A. Ashburn, “Physiotherapy Based on the Bobath Concept in Stroke Rehabilitation: A Survey The results from this investigation suggest that improvement within the UK,” Disability and Rehabilitation, vol. 23, no. 6, in functional and neuromechanical outcomes after 4 weeks of pp. 254–262, 2001. RE gait training can be achieved in adolescents and young [10] D. R. Louie and J. J. Eng, “Powered robotic exoskeletons in adults with chronic ABI. This study suggests that there could post-stroke rehabilitation of gait: a scoping review,” Journal be potential long-term effects of improved loading and of Neuro Engineering and Rehabilitation, vol. 13, no. 1, 2016. unloading, increased step length, and increased speed due [11] D. A. Winter, Biomechanics and Motor Control of Human to RE gait training. While the current results are promising, Gait: Normal, Elderly, and Pathological, Waterloo Biomechan- future studies with a larger sample would be required to fur- ics, Canada, 2009. ther understand the efficacy of the RE in adolescents and [12] M. T. Jahnke, S. Hesse, C. Schreiner, and K. H. Mauritz, young adults to confirm any training effect conclusively. “Dependences of Ground Reaction Force Parameters on Habitual Walking Speed in Hemiparetic Subjects,” Gait Pos- Data Availability ture, vol. 3, no. 1, pp. 3–12, 1995. [13] K. Karunakaran, N. Ehrenberg, J. Cheng, and K. J. Nolan, The data is currently unavailable due to Kessler Foundation “Effects of Robotic Exoskeleton Gait Training on an Adoles- IRB restrictions. cent with Brain Injury,” in IEEE 41h Annual Interna- tional Conference on Engineering in Medicine and Biology Society (EMBC), pp. 4445–4448, Berlin, Germany, Conflicts of Interest Germany, 2019. The authors have no conflict of interest. [14] K. J. Nolan and M. Yarossi, “Weight Transfer Analysis in Adults with Hemiplegia Using Ankle Foot Orthosis,” Pros- thetics and orthotics international, vol. 35, no. 1, pp. 45– Acknowledgments 53, 2017. We would like to acknowledge Danielle Nisenson, Kathleen [15] C. Chen, P. W. Hong, C. Chen et al., “Ground Reaction Force Chervin, and Brandon Ross for their assistance during this Patterns in Stroke Patients with Various Degrees of Motor Recovery Determined by Plantar Dynamic Analysis,” Chang study. Research supported by the New Jersey Health Founda- Gung medical journal, vol. 30, no. 1, 2007. tion PC5-18, Children’s Specialized Hospital, the Kessler [16] A. M. Wong, Y. C. Pei, W. H. Hong, C. Y. Chung, Y. C. Lau, Foundation, and the Reitman Foundation. and C. P. Chen, “Foot Contact Pattern Analysis in Hemiplegic Stroke Patients: An Implication for Neurologic Status Deter- References mination,” Archives of physical medicine and rehabilitation, vol. 85, no. 10, pp. 1625–1630, 2004. [1] R. A. Newton, “Balance abilities in individuals with moderate [17] J. Perry and J. Burnfield, “GAIT Normal and Pathological and severe traumatic brain injury,” Brain Injury, vol. 9, no. 5, Function,” Journal of Sports Science and Medicine, vol. 9, pp. 445–451, 2009. no. 2, p. 353, 2010. [2] G. Williams, M. E. Morris, A. Schache, and P. R. McCrory, [18] H. J. Dananberg, “Functional Hallux Limitus and its Relation- “Incidence of Gait Abnormalities after Traumatic Brain ship to Gait Efficiency,” Journal of the American Podiatric Injury,” Archives of physical medicine and rehabilitation, Medical Association, vol. 76, no. 11, pp. 648–652, 1986. vol. 90, no. 4, pp. 587–593, 2009. [19] P. A. Goldie, T. A. Matyas, O. M. Evans, M. Galea, and T. M. [3] K. M. Michael, J. K. Allen, and R. F. MacKo, “Reduced Ambu- Bach, “Maximum Voluntary Weight-Bearing by the Affected latory Activity after Stroke: The Role of Balance, Gait, and Car- and Unaffected Legs in Standing Following Stroke,” Clinical diovascular Fitness,” Archives of physical medicine and Biomechanics, vol. 11, no. 6, pp. 333–342, 1996. rehabilitation, vol. 86, no. 8, pp. 1552–1556, 2005. [20] A. Esquenazi, M. Talaty, and A. Jayaraman, “Powered exoskel- [4] M. A. Kemu, “Kinetics and kinematics of loading response in stroke patients (a review article),” Annals of King Edward Med- etons for walking assistance in persons with central nervous system injuries: a narrative review,” PM&R, vol. 9, no. 1, ical University, vol. 14, no. 4, 2008. pp. 46–62, 2017. [5] M. G. Benedetti, V. Agostini, M. Knaflitz, V. Gasparroni, M. Boschi, and R. Piperno, “Self-Reported Gait Unsteadiness [21] F. Molteni, G. Gasperini, G. Cannaviello, and E. Guanziroli, in Mildly Impaired Neurological Patients: An Objective “Exoskeleton and end-effector robots for upper and lower Assessment through Statistical Gait Analysis,” Journal of neu- limbs rehabilitation: narrative review,” PM&R, vol. 10, 9 Suppl roengineering and rehabilitation, vol. 9, no. 1, p. 64, 2012. 2, pp. S174–S188, 2018. 10 Applied Bionics and Biomechanics [38] R. S. Calabrò, A. Naro, M. Russo et al., “Shaping Neuroplasti- [22] H. Igo Krebs, N. Hogan, M. L. Aisen, and B. T. Volpe, “Robot- aided neurorehabilitation,” IEEE Transactions on Rehabilita- city by Using Powered Exoskeletons in Patients with Stroke: A tion Engineering, vol. 6, no. 1, pp. 75–87, 1998. Randomized Clinical Trial,” Journal of neuroengineering and rehabilitation, vol. 15, no. 1, p. 35, 2018. [23] A. M. Dollar and H. Herr, “Lower extremity exoskeletons and active orthoses: challenges and state-of-the-art,” IEEE Trans- [39] G. J. Androwis, R. Pilkar, A. Ramanujam, and K. J. Nolan, “Electromyography Assessment during Gait in a Robotic Exo- actions on Robotics, vol. 24, no. 1, pp. 144–158, 2008. skeleton for Acute Stroke,” Frontiers in neurology, vol. 9, 2018. [24] K. Anam and A. A. Al-Jumaily, “Active exoskeleton control systems: state of the art,” Procedia Engineering, vol. 41, pp. 988–994, 2012. [25] L. Dipietro, M. Ferraro, J. J. Palazzolo, H. I. Krebs, B. T. Volpe, and N. Hogan, “Customized interactive robotic treatment for stroke: EMG-triggered therapy,” IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. 13, no. 3, pp. 325– 334, 2005. [26] A. Jayaraman, S. Burt, and W. Z. Rymer, “Use of lower-limb robotics to enhance practice and participation in individuals with neurological conditions,” in Pediatric Physical Therapy, vol. 29, pp. S48–S56, 2017. [27] D. J. Reinkensmeyer, J. L. Emken, and S. C. Cramer, “Robotics, Motor Learning, and Neurologic Recovery,” Annual review of biomedical engineering, vol. 6, no. 1, pp. 497–525, 2004. [28] US F& DAdministration, Ekso User Manual, 2020, https:// www.accessdata.fda.gov/cdrh_docs/pdf20/K200574.pdf. [29] S. Winiarski and A. Rutkowska-Kucharska, “Estimated Ground Reaction Force in Normal and Pathological Gait,” Acta of Bioengineering & Biomechanics, vol. 11, no. 1, [30] D. A. Winter, Biomechanics and Motor Control of Human Movement, John Wiley & Sons., Hoboken, New Jersey, 2009. [31] K. Fujita, H. Hori, and Y. Kobayashi, “Contribution of Muscle Activity at Different Gait Phases for Improving Walking Per- formance in Chronic Stroke Patients with Hemiparesis,” Jour- nal of physical therapy science, vol. 30, no. 11, pp. 1381–1385, [32] T. S. Keller, A. M. Weisberger, J. L. Ray, S. S. Hasan, R. G. Shiavi, and D. M. Spengler, “Relationship between Vertical Ground Reaction Force and Speed during Walking, Slow Jog- ging, and Running,” Clinical biomechanics, vol. 11, no. 5, pp. 253–259, 1996. [33] S. R. Husain, A. Ramanujam, K. Momeni, and G. F. Forrest, “Effects of exoskeleton training intervention on net loading force in chronic spinal cord injury,” in Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, Honolulu, HI, USA, 2018. [34] A. Ramanujam, A. Spungen, P. Asselin et al., “Training response to longitudinal powered exoskeleton training for SCI,” in Biosystems and Biorobotics, 2017. [35] N. M. Salbach, K. O'Brien, D. Brooks et al., “Speed and dis- tance requirements for community ambulation: a systematic review,” Archives of Physical Medicine and Rehabilitation, vol. 95, no. 1, pp. 117–128.e11, 2014. [36] A. Rojek, A. Mika, Ł. Oleksy, A. Stolarczyk, and R. Kielnar, “Effects of Exoskeleton Gait Training on Balance, Load Distri- bution, and Functional Status in Stroke: A Randomized Con- trolled Trial,” Frontiers in Neurology, vol. 10, 2020. [37] F. Molteni, G. Gasperini, M. Gaffuri et al., “Wearable Robotic Exoskeleton for Overground Gait Training in Sub-Acute and Chronic Hemiparetic Stroke Patients: Preliminary Results,” European journal of physical and rehabilitation medicine, vol. 53, 2017.

Journal

Applied Bionics and BiomechanicsHindawi Publishing Corporation

Published: Oct 28, 2020

References