2.1.1. RT Group

Patients assigned to this group underwent gait training through the Gait Training and Evaluation System A3 (NX, China), which is a driven electromechanical gait robot that provides repeatable, high-intensity, and task-specific gait training. Automated exercise training was conducted on treadmills. Patients who did not participate in the assessment underwent supervised treatment with adjusted treadmill speed and weight support. This system involved dynamic and static weight loss systems, which can simulate real center of gravity changes when walking. As functions improve, the levels of weight support, treadmill speed, and guidance force are all adjusted to maintain the weak side of the knee extensor muscles during the standing position. The weight support level is gradually reduced from 50% to 0%, and the guiding force is reduced from 100% to 10% (by reducing the guiding force, which is used in both the standing and swinging phases, the patient is forced to use the hip and knee muscles to participate more actively in the gait process) [14, 15]. In addition, according to the tolerance of each patient, the treadmill speed (from 1.2 km/h) increased by 0.2 to 0.4 km/h per course of treatment, up to 2.6 km/h. The effective duration for each RT was 50 minutes.

2.1.2. PT Group

Conventional overground gait training is based on traditional neurodevelopmental therapy techniques. This therapy involved practicing sitting-standing balance, active transfer, sitting-standing, and intensive training for patients with sensorimotor disorders. With the improvement of physical functioning, the training of patients further increased in difficulty, including dynamic standing balance training, finally developing into functional gait training, while continuing to carry out intensive training [16].

Patients were assigned to this group for ground gait training (effective time of 50 minutes per lesson), aimed at improving posture control during gait, weight transfer, standing phase, free swing phase stability, heel full contact, and gait mode. The same trained therapist treated all patients in this group and standardized the performance of each exercise according to the patient’s skills (i.e., ability to participate in a progressive and more active manner during gait) and tolerance intensity, as previously described for the RT group.

3.1. Baseline

At the baseline assessment, no significant differences were observed between the two groups in terms of age (P = 0.14), stroke onset time (P = 0.47), FMA scores (P = 0.06), and TUG scores (P = 0.17). The demographic and clinical characteristics of patients are shown in Tables ​Tables11 and ​and22.

3.2. Outcome

Thus, the final analyses included 54 patients: 27 in the RT group and 27 in the PT group. Age, weeks poststroke, sex, side of stroke, and stroke type did not differ significantly between the two groups (see Table 1). We measured improvement by calculating the difference between the baseline and 2-week scores of each group. Because the data were not normally distributed, the Mann–Whitney U test was used to compare baseline and posttraining measurements between the two groups. There were no significant differences between groups in any outcome measurements before treatment.

After 14 training sessions, both groups showed significant improvements in at least one outcome measure. Moreover, the PT group exhibited a significantly greater performance improvement (see Table 2). Regarding FMA and TUG scores, the comparison of scores before and after 2 weeks of training revealed significant differences within the PT group (P < 0.01) (see Table 2) and significant differences in the RT group (FMA, P = 0.02), but the results of TUG (P = 0.28) exhibited no difference. The comparison between groups showed that there was no significant difference between the two groups in FMA scores (P = 0.26) or TUG scores (P = 0.97).

Regarding the time parameter gait analysis, in the intragroup comparison, there were no significant differences before and after each part of the two groups affected side (P > 0.05). In the intragroup comparison of the contralateral swing phase, the RT group was statistically significant (P = 0.01). In the symmetry of both sides of the lower limbs before and after two weeks of training in the standing period and swing period, the RT group was statistically significant in the intragroup analysis (P = 0.04). In addition, the stance phase, swing phase, and symmetry ratio of the less affected side and the affected side were not significant within and between groups (P > 0.05) (see Figure 2).

Figure 2

The blank bar represents the PT group, the diagonal bar represents the RT group, the light bar represents before treatment, and the darker bar represents after treatment. ∗P < 0.05.

Regarding the space parameter gait analysis, before and after 2 weeks of training, there was a significant difference in gait width on the affected side (P = 0.02) in the PT group. In the RT group, the affected side exhibited significant differences in walking velocity (P = 0.03), toe out angle (P = 0.01), and stride length (P = 0.03). However, after 14 days of training, the two groups did not exhibit any significant improvement in cadence. Except for the significant statistical difference in toe out angle (P = 0.002), no significant differences were revealed in the comparison between groups.

4. Discussion

The main purpose of this randomized controlled trial was to compare the effects of robot-assisted gait training (RT group) and conventional ground gait training (PT group) for early stroke patients with gait disorder. The current findings revealed that, compared with conventional ground gait training (PT group), gait training with the A3 robot using NX had several key advantages for improving motor function.

Several previous studies have reported that robotic gait training combined with physical therapy after stroke increased the likelihood of achieving independent walking compared with gait training without these devices, and people receiving this intervention in the first 2 months after stroke and those who could not walk were found to benefit the most [19, 20]. Our initial hypothesis was that robot-assisted gait training would be more effective than traditional ground gait training in improving athletic ability, by providing accurate and symmetrical walking patterns to regulate patients’ walking. In addition, we predicted that early robot-assisted training after stroke (i.e., dynamic regulation from the weight loss system, real-time adjustment of guidance force, and active and passive training at any time) would be more beneficial than traditional training based on information presented in clear language. Furthermore, we also speculated that gait training with the A3 robot in an upright position would activate the musculoskeletal and cerebrovascular systems through repeated and precise walking posture input, thereby alleviating spastic hypertonia and hyperreflexia and promoting early recovery from stroke.

The current findings did not fully confirm our initial hypotheses. FMA scores revealed that both groups showed significant improvements. In addition, in the early phase, the use of the robotic device to train the spatial parameters of gait led to significantly better performance than traditional ground rehabilitation training. After robot-assisted gait training, patients may not have been able to implement standardized gait quickly and skillfully, and patients’ time and space parameters were slightly higher than before training (although this difference was not significant, P > 0.05), with no significant difference in TUG scores before and after training (P = 0.28). However, regardless of the method, 2 weeks of continuous training did not change the time parameters in patients’ gait or step frequency in the space parameters.

The current findings are consistent with some previous reports, supporting the notion that the role of electromechanical/robot equipment is still unclear [10]. Some previous studies’ research has suggested that robotic gait training could play an early role in neurorehabilitation, providing correct sensory input as the premise of neural plasticity and the basis of motor learning, which is essential for achieving appropriate motor output [21]. Patients who received a combination of electrically assisted gait training and physical therapy after stroke were more likely to achieve independent walking compared with those who received only conventional gait training, especially in the first 3 months after stroke [7, 14]. In addition, some studies have shown that relying on robot training can improve walking of patients after stroke. In a study by Kim et al., 48 patients within 1 year of illness were divided into a robot-assisted treatment group (0.5 hours of robot training + 1 hour of physical therapy) and a conventional treatment group (1.5 hours of physical therapy), with both groups receiving 1.5 hours of treatment per day. Compared with traditional physical therapy alone, the results revealed that combining robotic devices with physical therapy was superior to conventional therapy in terms of autonomy and balance [22].

However, Mayr and colleagues conducted a study of 66 adult patients with an average of 5 weeks after stroke to evaluate the impact of two groups receiving 8 weeks of inpatient rehabilitation treatment focused on gait ability and gait rehabilitation (robot-assisted gait training and traditional ground gait training). It was reported that, although it took time and energy to achieve beneficial effects of gait training exercise, both methods improved gait function [15]. Similarly, Duncan et al. examined the effects of early exercise training (2 months after stroke onset), late exercise training (6 months after stroke onset), and a home exercise plan (2 months after stroke onset) to study weight-supported running after stroke, including the optimal timing and effectiveness of the mechanical rehabilitation intervention. It was found that, among 408 adult patients with stroke (2 months after stroke), exercise training, including the use of treadmill training for weight support, was no better than exercise therapy performed by a physical therapist at home [8]. Hidler and colleagues proposed a multicenter RCT study that included 72 adult patients less than 6 months after the onset of stroke. The authors report that in individuals with moderate to severe gait disorder after a subacute unilateral stroke, the use of traditional rehabilitation strategies can achieve greater speed and distance on the ground than robot-assisted gait training (using Lokomat devices) [9]. In our study, it can be seen from the comparison between the groups that, except for the significant statistical difference in toe out angle, in fact, the treatment effect of the PT group is similar to that of the RT group in most aspects. Especially in terms of gait width, after 2 weeks of PT training, the intragroup comparison is significant (P = 0.02). This reminds us that in rehabilitation training centers without robot training conditions, gait training with conventional overground gait training can also achieve a certain therapeutic effect.

In terms of clinical implications, the current findings tentatively suggest that, for clinical gait training for early stroke, when the patient’s gait width is problematic, conventional overground gait training should be chosen; in contrast, when the patient’s space parameters (step length, pace, and toe angle) or time parameters (stance phase symmetry ratio) reveal a gait problem, choosing robot-assisted gait training may be more appropriate. However, the main limitation of the current randomized controlled trial was the relatively short training time (2 weeks), limiting the conclusions that can be drawn from our findings. It is possible that training differences between the two methods would be revealed after 4 weeks. A second limitation is related to the study population. The current study was conducted with patients with subacute strokes of different levels of severity, and we were not able to distinguish between spontaneous rehabilitation (means spontaneous recovery of the body) and therapeutic rehabilitation. The selection period (8 weeks) from the onset of stroke was relatively long, possibly involving an excessive number of different spontaneous evolution curves and individual resistance to (training) stress. Another important limitation is the lack of long-term measurement points (e.g., 6 months or more and ideally 1 year). Moreover, starting treatment (i.e., RT) early may not result in a measurable difference in short-term results, even if it does achieve a difference in long-term results.

5. Conclusion

This preliminary study shows that both A3 robot-assisted gait training and conventional ground gait training can partially improve the walking ability of stroke patients within 2 weeks.

Acknowledgments