Participants
Participants were recruited from community-dwelling elderly people in Seoul, Korea. All participants were eligible for inclusion if they met the following requirements: they were over the age of 65, they had sufficient cognition to follow simple instructions and to understand the content and purpose of the study (based on the Korean version of the Montreal Cognitive Assessment of ≥ 23 points), and they did not have a musculoskeletal condition that could potentially affect the ability to walk safely. Participants were excluded if they had severe heart disease or uncontrolled hypertension and pain or if they had any neurological disease that might interfere with the study. Ethics approval was granted from Samsung Medical Center institutional Ethics Committee, and written informed consent was obtained from all participants prior to the study.
Design of the GEMS as a wearable hip-assist robot
The GEMS is developed to enhance gait function and to increase the community locomotor interaction in patients with gait disorders and elderly people. The GEMS is a lightweight (2.8 kg), slim, comfortable, and powerful active assist robot. As shown in Figure 1., the GEMS consist of a pair of actuators for transmit assistance torque of the both hip joints, a hip brace that fits around the waist, a pair of thigh carbon frames that carry on assistance torque from the actuators to the thighs. It also had fabric belts at the ends of the thigh frames. The assist torque element in GEMS consist of angular sensors and actuators on both hip joints. The hip joint angle is given to the controller, then PSAO (Particularly Shaped Adaptive Oscillator) estimates the gait phase or percent gait cycle. The output torque is then calculated from the torque look-up table based on the actual gait phase. The details information of the GEMS's assistant algorithm are described in our previous paper [21, 22]. In addition, a research administrator used a tablet PC (Galaxy Tab 3 8.0 from Samsung Electronics Co., Ltd., Korea with the Android 4.2 OS) to users’ weights and assistance levels for experiments and to log the data from GEMS at the rate of 100 Hz.
Experimental design
A randomized, controlled trial was performed to test the effectiveness of a gait training using a newly developed wearable hip-assist robot with community-dwelling elderly adults. All participants were recruited by way of an advertising poster with information about enrollment, study objectives, and exclusion criteria. The eligible participants were randomly placed in either the experimental group (gait training with wearable hip-assist robot) or the control group (gait training without wearable hip-assist robot) by a research administrator using a random number table after baseline assessment. All participants were assigned a code number. The intervention was administered by two physiotherapists. The study flow diagram is outlined in Figure 2.
Intervention protocol
Experimental participants received an intensive gait training program with a total of 10 sessions during three sessions per week for four weeks involving five sessions of treadmill gait training with GEMS and five sessions of over-ground gait training with GEMS. The goal of the gait training was to facilitate improvements in gait functions (walking speed, muscle efforts of trunk and lower extremity) and cardiopulmonary metabolic energy consumption. All participants were randomly assigned to treadmill or over-ground gait training by a random number table. For treadmill gait training, the participants wore a harness without any body weight support and used handrails to prevent falls. In addition, the first treadmill training session was used to adjust the device properly to the individual participant and allow the participants to get comfortable with GEMS. In contrast, the control group received treadmill gait training at their most comfortable speed without GEMS. We gradually increased treadmill speed (normally up to 3 km/h) in order to challenge the participant to walk as actively as possible with GEMS during five treadmill gait training sessions. At the final session, the experimental group reported a faster comfortable speed on the treadmill than the control group. Also, during the five sessions of over-ground gait training, participants performed training at their most comfortable speed with (experimental group) or without (control group) GEMS in the corridor in front of our laboratory.
According to previous research based, similar to our study, in the training protocol, the average time per session was 45 min, divided into a 5 min warm-up, 35 min intensive gait training (including 5 min resting time) and 5 min cool-down [23, 24]. The structure of the intervention was customized to each participant’s physical function level.
Outcome measures
Gait function as a spatio-temporal parameter and muscle effort while walking were the primary outcomes of this study. Spatio-temporal parameters (gait velocity, cadence, stride length and step width) were measured using a 3D motion capture system with six infrared cameras (Motion Analysis Corporation, Santa Rosa, CA, USA). In addition, muscle efforts while walking were measured using a 12-channel wireless surface electromyography (sEMG, Noraxon Inc., Scottsdale, AZ, USA). sEMG signals were registered via 12-channel wireless electromyography using 10 mm 3MTM Ag/AgCl surface electrodes. The electrodes were positioned on the participants’ right sides on the rectus abdominis (RA), external oblique (EO), erector spinae of the lower back (ES), hip flexor (HF), gluteus maximus (GMAX), gluteus medius (GMED), rectus femoris (RF), vastus medialis (VM), adductor longus (AL), biceps femoris (BF), tibialis anterior (TA), and the medial of gastrocnemius (GCM) muscles in accordance with the recommendations of the sEMG for the Non-Invasive Assessment of Muscles Project (SENIAM) [25]. Besides, foot-switch sensors were placed on participant plantar surface of the right toe and heel. The signals from the foot-switch sensors recorded data on the stance and swing phases during gait. The maximum voluntary contraction (MVC) data of all muscles for each subject was measured according to method of the SENIAM before performing the gait assessment [25]. After measuring the MVC, the EMG data of each muscle collected while the subject walked was normalized based on the walking cycle (%MVC), combined with the MVC data. The EMG signals were amplified and filtered (sampled at 1000 Hz with bandpass filtered between 10 Hz and 350 Hz) and full-wave rectified with Noraxon software (MyoResearch XP Master Edition).
Cardiopulmonary metabolic energy consumption was the secondary outcome. This assessment was performed using a portable cardiopulmonary metabolic system (Cosmed K4B2, Rome, Italy). This system was worn on the back, to measure breath-by-breath metabolic cost. The flow meter was calibrated using a 3-L syringe. Energy consumption variables collected included oxygen consumption (ml/min/kg) and an aerobic energy expenditure measurement (EEm; Kcal/min). Cardiopulmonary metabolic energy consumption measurements consisted of a resting and gait test on a treadmill. The resting condition was obtained during a standing position for three minutes prior to the gait trial. Following this, participants walked continuously for six minutes on the treadmill at their most comfortable speed without GEMS assist. The participants were given specific instructions not to talk or laugh during measure. Particular care was taken that each participant walked on the treadmill while wearing a safety harness and that each participant walked comfortably without using handrails. All participants were performed clinical measurements without GEMS at before the intervention and immediately after the ten sessions intervention.
Statistical Analysis
All statistical analysis were undertaken using SPSS version 22.0 (IBM, Armonk, N.Y., USA) and the level of significance was set at 0.05. The Shapiro-Wilk test was used to confirm that all outcome variables were normally distributed. The independent t-test for continuous parameters, Mann-Whitney U test for ordinal parameters, and x2 test for categorical parameters were used to compare the baseline characteristics of the participants in both groups. For measures of dependent parameters, a repeated measures analysis of variance with mixed design (within-subjects factor = time, between subjects factor = group) was used to compare each outcome variable between two time points (pre- and post-test). Means, SDs and 95% CIs were provided to depict the change within each group during the course of the study and the training effect.