Evaluating a pediatric exoskeleton to improve walking function in children with movement disorders

Clinical Center

Investigators

Linked publications, trials & patents

Abstract

At the time this project was initiated, no pediatric exoskeleton designed specifically for children with crouch gait (or cerebral palsy) existed. The NIH research team spent several years developing new robotic technology to fill this gap. We built a prototype exoskeleton, which we called the extension assist knee ankle foot orthosis (EA-KAFO), from the ground up to implement a novel control scheme designed specifically to improve knee extension while maintaining user muscle activity during walking. We then designed an observational trial to assess the biomechanical and neuromuscular effects of the exoskeleton on gait biomechanics. The initial study design included six total visits with three data collection time points, one at the beginning of the study (baseline), one immediately after the participant was able to walk independently with the exoskeleton (initial assessment) and one at the final visit (final assessment). The primary outcome measure was peak knee extension during midstance during walking, with secondary outcome measures including gait speed, step length, knee extensor and flexor muscle activity, and knee joint moment in the sagittal plane. We completed the first clinical cohort study of this novel EA-KAFO involving seven participants who completed the full set of six visits. Three different modes of assistance were implemented in this first study, during which knee extension assistance was provided during stance phase only, during swing phase only, or during both stance and swing phase. All participants were able to achieve independent overground walking in the exoskeleton without a mobility aid or therapist assistance, with 6/7 able to do this within the first practice session. At the group level the stance & swing assist condition was the most effective at alleviating crouch gait, with mean stance phase knee extension significantly increased by 13.3 degrees (p=0.024) and 5.8 degrees (p=0.039) for the more- and less-affected limbs, respectively. Importantly, these improvements occurred without a significant reduction in vastus lateralis (knee extensor) muscle activity, suggesting the exoskeleton may be a viable long-term rehabilitation strategy. We performed a detailed analysis of the exoskeleton effects on gait biomechanics over the course of the visits (i.e., between the first and final assessments) and also on the kinetics and muscle activity underlying the improvements in knee extension during exoskeleton use. Improvement in knee extension with the exoskeleton accrued with practice as on average knee extension improved by 5.3 degrees (p=0.003) between assessments, indicating a progressive reduction in crouch. Further, step length (p=0.025) and gait speed (p=0.023) were significantly increased between first and final assessments with the exoskeleton. We also found that the exoskeleton reduced excessive stance-phase knee extensor moment by 35% in early stance and 76% in late stance, on average, reducing the excessive burden on extensor muscles during walking in those with crouch. These reductions translated to a 12 degree increase in hip extension. Based on the outcomes of this initial cohort, we undertook significant design revisions to the robotic exoskeleton to refine the mechanical and electrical hardware as well as the software design that governs the interaction between the robot and wearer. These updates resulted in a second-generation device that was finalized in 2019, which had a reduced footprint, greater assistive torque capability (to enable use in a broader range of children with less functional mobility), and a refined control system that segments the gait cycle into up to 5 discrete states and provides tunable assistance within each as well as an adaptive control mode that provides assistance proportional to the instantaneous knee joint moment. This second-generation device integrated a newly developed controllable surface electrical stimulator, which allowed for up to 4 channels of surface electrical stimulation to be provided in synchrony with exoskeleton assistance. These design revisions and novel assistance modes were evaluated in 2 healthy volunteers who each completed a single session with the exoskeleton; data from these individuals validated the design. In parallel with these design revisions and validation testing, we amended the protocol to a) expand the population of users to include other pediatric movement disorders which exhibit reduced knee extension due to muscle weakness, including spina bifida, muscular dystrophy, and incomplete spinal cord injury; b) expand the total number of visits from 6 to 10 to enable further evaluation of the accumulating benefit of knee extension assistance across visits. One participant with CP was recruited to evaluate the second-generation device over 10 visits; 8/10 visits were completed before the protocol was paused due to the COVID-19 pandemic. Nevertheless, we were able to provide initial validation of the next generation control systems in the expanded target population with reduced mobility (GMFCS III) compared to the first cohort of individuals who were GMFCS I/II. In 2019, a Cooperative Research and Development Agreement (CRADA) with Bionic Power, a company based in Vancouver, Canada, was established to bring this technology to market. This CRADA involved new (3rd) generation hardware utilizing the Bionic Power Agilik (R) actuators combined with the NIH custom KAFO architecture, sensors, user interface, and embedded control strategy. This new device, termed the NIH-Agilik, further expanded the control capabilities by allowing bidirectional application of torque with low latency. This created a new paradigm for gait training using the robotic exoskeleton, whereby controllable resistance to knee extension (i.e., a flexion torque) could be applied during the gait cycle. The purpose of this new mode would be to provide targeted, functional resistance exercise to the user that can result in improved knee extension after removal of the applied resistance (and exoskeleton). This new strategy required submission of a second FDA risk assessment (beyond the initial risk assessment submitted with the original protocol). The FDA determined the new approach to be non-significant risk, and the protocol was amended to include this new operational mode for evaluation. Recruitment resumed after the pandemic related pause in December 2020. To date, we have completed the 10-visit study with the NIH-Agilik in three participants (two with CP, one with spina bifida). The improvements in peak knee extension in the Assist mode have been larger, on average, than those observed with the first-generation exoskeleton. Additionally, each participant has been able to walk overground with controllable resistance applied during stance and/or swing phase, suggesting that this approach to functional resistance training is feasible in the target clinical population. The cohort study evaluating these different modes of exoskeleton operation is ongoing with an estimated cohort size of n=10 to determine which of the new operational modes is most effective.