We found that robotized KAFO-assisted gait training could be performed safely using a safety assessment to prevent risk. In patients with chronic stroke, long-term robotized KAFO-assisted gait training may improve genu recurvatum during overground gait without robotic device, motor function of the paralyzed lower limb, efficacy of gait, and gait pattern in the patient with chronic stroke. We suggest that robotized KAFO-assisted gait training has a therapeutic effect on genu recurvatum. This is the first report using an exoskeleton robot regulating the knee and ankle joints simultaneously. This has the potential to improve genu recurvatum during gait in patients with stroke.
The safety of robotized KAFO-assisted gait training
Robotized KAFO-assisted gait training was safely performed in both healthy participants and patients with stroke. Previous studies have assessed the safety of robot-assisted gait training in patients with neurological disorders using a self-report questionnaire [26], assessment of pain [27], recording of adverse events [26–29], and assessment using the U.S. Food and Drug Administration’s list of known and unforeseen adverse events [30]. Before the training, we listed the known and unforeseen risks and adverse events that both the robot and participants have during robot wearing, training, and after taking off the robot. We developed a safety assessment protocol, including a checklist of risks and questions to ask participants. As a result, we were able to detect the pain caused by robot fitting and prevent adverse events. Using a safety assessment may make it possible to safely perform robot-assisted gait training. It is necessary to analyze many differences in robotic devices and diseases; however, we suggest that this safety assessment can be used for gait rehabilitation using an exoskeletal robot in general. Detailed investigation including kinematic data is therefore necessary in the future.
The effect of 10 days intervention on genu recurvatum (Experiment 2)
The time from the onset of stroke in patient C in Experiment 2 (the patient received 10 days intervention) was 3 years and 8 months. Genu recurvatum during gait is habitual, so the patient had difficulty to prevent genu recurvatum during gait by herself before the intervention. After 10 days of the intervention, however, the patient achieved overground gait without genu recurvatum, even without robotized KAFO. Robotized KAFO-assisted training is thought to stimulate the learning of joint movements without genu recurvatum.
In a previous report on robotic intervention for genu recurvatum, the fifteen-days intervention using a robotized knee orthosis with knee joint assist showed improvement in maximum flexion of knee angle during gait and decrease in knee extension moment, among others, but could not prevent the appearance of knee hyperextension during both stance and swing phases [6]. The causes of genu recurvatum are not only the knee joint, but also movement disorders of the hip and ankle joints [2, 5, 7]. Therefore, to change this disorder during walking, it may be effective to assist both the knee and ankle joints.
The barefoot walking of the patient on the overground before the intervention resulted in less angular changes between the knee and ankle on the affected side and loss of coordination between the knee and ankle. The training focused on heel contact with the knee in a slightly flexed position, promotion of lower leg forward tilting after heel contact, and relearning of push-off in the pre-swing phase. The function of the device to independently adjust the timing and power of the assist in the four directions of joint motion may have enabled tailor-made assistance for her gait pattern. Furthermore, real-time feedback of the knee joint angle was also performed using a monitor so that the patient could be aware of heel contact with the knee in a slightly flexed position. As a result, the patient acquired a gait without genu recurvatum, even while she walks without robotized KAFO. We suggest that re-learning of joint motion with an exoskeleton robot assistance and motor learning with angle feedback may improve genu recurvatum. Considering that proprioceptive training using videographic observation improved genu recurvatum [14], motor learning with joint angle feedback also plays an important role in the improvement of genu recurvatum.
In Experiment 2, the FMA-LE score also showed improvement in E-I (reflex activities), E-II (volitional movement with dynamic flexor synergies), E-IV (volitional ankle dorsiflexion), and F (coordination and speed). Voluntary ankle dorsiflexion movement improved despite the chronic phase. Previous studies in patients with acute and sub-acute stroke have reported improvement in FMA-LE with robot-assisted gait training [31–36]. Although some studies have shown similar improvements in FMA-LE after robot-assisted gait training in patients with chronic stroke [37, 38], the number of sessions or duration of intervention were longer than in our intervention. The robotized KAFO can independently adjust the flexion/extension assistance of the knee joint and plantar flexion/dorsiflexion assistance of the ankle joint, as required. Furthermore, the MAS scores of the quadriceps, hamstrings, and triceps surae muscles improved after this intervention. The causes of spasticity include neural changes, muscle atrophy, and muscle contractures [39–42]. Increased motion of the knee joint during gait may have resulted in stretching and shortening of the thigh muscles and affected the viscoelasticity of the muscle-tendon complex. Spasticity in the quadriceps or triceps surae muscle is thought to be part of the cause of genu recurvatum [2, 5, 7]. Reduced spasticity of these muscles may have led to reduced extensor synergy and improved genu recurvatum.
In addition to the qualitative component of lower-limb movement, it is important to not lower the efficiency of gait. The patient did not show a decrease in walking speed after training, while controlling for genu recurvatum. Rather, the results of a 6-min barefoot walk test showed improved walking efficiency over long distances. After the intervention, flexor synergy (motion of the hip and knee joints) and voluntary ankle dorsiflexion on the affected side improved. Furthermore, the EMG activation pattern of the knee extensors on the affected side changed. Before the intervention, the patient showed low VM activity in the early stance phase due to genu recurvatum. After the intervention, VM activity increased at the same time as heel contact did. Simultaneously, there was a decrease in GM activity. It is generally known that the GM is active in maintaining forward movement of the center of gravity against the impact of the ground during the initial contact of walking [43]. The decrease in GM activity at initial contact in this patient may be related to the reduced impact of contact with the ground owing to regaining initial contact with a slightly flexed knee position. Since this was a single case study without control intervention, the training effect of treadmill walking should be considered, but the changes in leg motor function and muscle activation pattern on the affected side may contribute to improved walking efficacy. Further studies are needed to investigate the mechanisms that support the changes in gait and muscle activity.
This is the first study in which an exoskeletal robot showed after-effects on the improvement of genu recurvatum. Therefore, it is necessary to increase the number of participants in order to establish more stringent evidence in future.