Several clinical trials of robot-assisted gait training have been reported in neurology patients. ^{1– 4} In particular, the Hybrid Assistive Limb (HAL) gait training has resulted in improved gait and balance performance. ^{1, 3, 5} The HAL exoskeleton is completely driven by the patient’s own muscle activation, which is detected by surface electrodes on key lower extremity muscle groups. Thus, this self-initiated robotic-assisted movement has the potential to induce a somatosensory feedback-loop that enhances neural plasticity and locomotor function. ⁶ Although many previous studies have reported functional gains due to the HAL intervention (e.g., gait and balance), ^{1, 3, 5, 7} there are no studies that report its effects on neuromuscular activation likely to accompany functional gains. Understanding how HAL training changes neuromuscular activation patterns, will help in the evaluation and use of HAL training across various populations.

Multi-channel surface electromyography (SEMG) is a recently developed techniques used to evaluate single motor unit firing behavior and whole muscle EMG patterns of activity by using multiple electrodes arranged in a two-dimensional plane. ^{8– 10} This technique provides data on the spatial distribution of SEMG within a muscle. Previous studies have demonstrated that the spatial distribution of SEMG is altered by contraction levels or fatigue during isometric contraction. ^{11, 12} This phenomenon has been explained by a spatial inhomogeneity in the location of different types of muscle fibers ¹³ and a clustering of muscle fibers innervated by one motor unit in a limited territory. ^{14– 16} Previous studies have demonstrated that alterations in the multi-channel SEMG spatial distribution pattern can be explained by the physiological phenomenon of motor unit recruitment and suggested that the multi-channel SEMG spatial distribution pattern could be used to study changes in motor unit recruitment patterns. ^{15, 17} In addition, previous studies have used metrices such as the inhomogeneity of the spatial distribution pattern of SEMG (EMG variables; modified entropy and coefficient of variation (CoV) of root mean square (RMS) ^{11, 18} to quantify neuromuscular fatigue and the influence of force generation. The assessment of motor unit recruitment is very important for determining the effect of neurorehabilitation on neuroplasticity in patients with several types of neurological conditions. Thus, we considered that this technique (i.e., multi-channel SEMG) could be employed as a new assessment tool to evaluate changes in muscle activation as a metric of neuroplasticity in response to HAL training.

Individuals with spinal muscular atrophy (SMA) type III present with progressive proximal weakness of the legs more than arms, and the leg weakness is so profound that it can necessitate the long term need of a wheelchair for mobility. ¹⁹ Rehabilitation is one treatment for SMA type III. The main objectives for rehabilitation of individuals with SMA type III who are still able to walk are to maintain, restore or promote function, mobility, and adequate joint range, and improve balance and endurance. ²⁰ HAL gait training has been an insurance supported therapy since January 2016, and SMA, amyotrophic lateral sclerosis, spinal bulbar atrophy, Charcot-Marie-Tooth disease, inclusion body myositis, distal myopathy, and congenital myopathy are targeted disease populations. Although approved by insurance, there is limited information on how the effectiveness of HAL training compares to other forms of therapy. Specifically, there is no report which compares HAL and conventional treadmill gait training on neuromuscular activation in SMA patients that likely accompany improvements in walking function. Therefore, the aim of this case report to quantify the effects of a single session of HAL gait training on SEMG spatial distribution patterns in a person with spinal muscular atrophy type III. We hypothesized that HAL gait training would induce more heterogenous muscle activation reflecting more varied motor unit recruitment than treadmill gait training.

As a neuromuscular activation assessment, multi-channel SEMG was performed on the patient before and after both gait training. Figure 1 illustrates the representative multi-channel SEMG amplitude color maps for treadmill and HAL gait training. We normalized the CoV of force and RMS estimates to the values obtained pre-intervention for each torque level. Differences were observed in the multi-channel SEMG spatial distribution pattern at each contraction torque between HAL and treadmill gait training. There was a greater decrease ΔCoV force with the HAL gait training than for treadmill gait training ( Figure 2A). The increase in the magnitude of muscle activation from pre-intervention to post-intervention was greater for HAL gait training than for treadmill gait training ( Figure 2B). The CoV of the RMS and modified entropy at each torque in both gait trainings are shown in Figure 3. For HAL training, both the CoV of the RMS were higher and the modified entropy was lower as compared to treadmill gait training ( Figure 2C and D). No adverse events were observed with either intervention. He did not experience worsening of the symptoms during each gait training. Figure 1.

(upper panel) treadmill gait training, (lower panel) HAL gait training. Differences in the multi-channel surface electromyography spatial distribution patterns at each torque level were observed between the treadmill and HAL gait training in these representative data.

We applied multichannel SEMG as a neuromuscular activation assessment of HAL training. In addition to the overall increase in EMG activity, the HAL gait training resulted in decrease force fluctuation, and greater heterogeneity in spatial muscle distribution pattern from pre-intervention to post-intervention than for the treadmill gait training.

We used two parameters to quantify the multi-channel SEMG spatial distribution pattern: modified entropy and the CoV of the RMS. A decrease in the modified entropy and an increase in the CoV of the RMS are consistent with increased heterogeneity in the multi-channel SEMG spatial distribution pattern within the electrode grid. ²⁵ The SEMG spatial distribution pattern can be altered by contraction levels or fatigue during isometric contraction. ^{11, 12} Measures of the CoV of the RMS and modified entropy provide insight into how the nervous system regulates muscle activation across the muscle irrespective of the magnitude. ²⁸ This phenomenon has been explained by spatial heterogeneity in the location of different types of muscle fibers ¹³ and a clustering of muscle fibers innervated by one motor unit in a limited territory, ^{14, 15} and a previous study reported that alterations in the multi-channel SEMG spatial distribution pattern can be explained by the physiological phenomenon of motor unit recruitment patterns. ¹⁷ The results of the present report for the HAL training intervention showed that the CoV of the RMS was higher, and the modified entropy was lower than for treadmill gait training. These findings suggest that spatial motor unit activity patterns were enhanced by HAL training. Changes in the heterogeneity of SEMG were accompanied by a greater decreased in the Δ CoV force for the HAL gait training as compared to the treadmill gait training. This finding suggests that the participant was able to perform the submaximal ramp up task smoothly by altered spatial motor unit activity patterns.

The HAL gait training offers the possibility to monitor muscle contractions via SEMG at the extensor and flexor muscle region of the lower extremities. ^{5, 29} The voluntary drive and normalized motion assistance provided by the external device likely form the foundation for a proprioceptive feedback loop for patients with lesions involving the sensory pathway and not facilitated as well with treadmill training. The neural activity and repeated execution of specific tasks promote learning and lead to the reinstatement or restructuring of appropriate proprioceptive feedback. ^{30, 31} Here, we are the first to show that HAL gait training induced an alteration muscle activation patterns due to a single session of HAL training. Moreover, based on our findings, HAL gait training may be more effective in evoking changes in neuromuscular activation than conventional gait training. Future studies will examine the relationship between changes in gait and muscle activation using the HAL intervention across multiple sessions.

This report has several limitations. First, report included only one patient with spinal muscular atrophy type III. Future large-sample studies and studies of patients with various neurological conditions (e.g., amyotrophic lateral sclerosis, Parkinson’s disease, and stroke) are needed to clearly understand how generalizable the effects of HAL gait training on neuroplasticity. Second, this report only included data following a single session. Future studies will examine how factors such as dose, frequency, and intensity of HAL training affect changes in function and neuromuscular activation.

The present case report showed that HAL gait training was more effective in changing metrics of neuromuscular activation of the knee extensors than treadmill gait training despite similar training durations. These data suggest that the modes of gait training which incorporate both voluntary activation and guided movement of the legs may facilitate recovery of function.

Written informed consent for publication of their clinical details and/or clinical images was obtained from the patient/parent/guardian/relative of the patient.