Hemiparetic gait impairment is a hallmark sequela in stroke victims and affects ambulation and quality of life [1]. Due to advanced medical technology and care, the mortality rate from stroke has significantly decreased. However, as many as 80% of stroke survivors suffer from locomotor dysfunction, which is characterized as asymmetrical step length, slow velocity, and altered biomechanical alignment [23].

Hemiparetic gait kinematics encompass excessive ankle plantar flexion during the swing phase and knee hyperextension and hip flexion during the stance phase, whereas kinetically decreased and asymmetrical ground reaction force is manifested during the push up phase. Recent stroke rehabilitation evidence suggests that a large number of repetitions (> 600) [4] and accurate sensorimotor feedback (e.g., kinematics and kinetics) are necessary to produce neuroplastic changes and associated locomotor functional recovery in stroke patients [5].

To restore hemiparetic gait, 2 innovative robot-assisted gait training paradigms utilizing the end-effector and exoskeleton robot types have recently been adopted to provide an ample number of repetitions with precise kinematics and kinetic sensorimotor feedback. In fact, the cumulative robotics studies demonstrated promising effects when robot-assisted gait training was employed using either end-effector or exoskeleton types in conjunction with conventional gait training. Conventional gait training often poses inherent issues, as it is labor intensive, costly, and does not provide accurate sensorimotor feedback.

Hence, the present study underwent an extensive review of the literature to determine the best clinical evidence of the most commonly utilized robot-assisted gait training paradigms (end-effector and exoskeleton types) in stroke gait rehabilitation.

End-effector devices work by applying mechanical forces to the distal segments of limbs. In end-effector devices, a subject's feet are placed on foot-plates, whose trajectories simulate the stance and swing phases during gait training [6]. End-effector type robots offer the advantage of easy setup, but suffer from limited control of the proximal joints of the limb, which could result in abnormal movement patterns. Examples of end-effector devices are the Gate Trainer GT1 (Reha-Stim, Berlin, Germany) and the G-EO-System (Reha Technology AG, Olten, Switzerland). In contrast, exoskeleton-type robotic devices have robot axes aligned with the anatomical axes of the wearer. These robots provide direct control over individual joints, which can minimize abnormal posture or movement. Construction of exoskeleton-type devices is more complex and more expensive than that of the end-effector type. Exoskeleton devices are outfitted with programmable drives or passive elements that move the knees and hips during the phases of gait [6]. Examples of the exoskeleton type of device are the Walkbot (P&S Mechanics, Seoul, Korea) and the Lokomat (Hocoma AG, Zurich, Switzerland).

In this manuscript, we used a Cochran search using keywords, including robot-assisted gait training, end-effector type, and exoskeleton type, to summarize the past 10 years of research concerning both end-effector and exoskeleton types of gait robot devices. The current status of robot-assisted therapy in stroke rehabilitation was discussed.

The results of 10 randomized controlled trials that investigated end-effector robot gait training in stroke patients are summarized in Table 1. Three studies compared therapeutic effects on motor recovery, gait function, and balance between a combination of the robot-assisted gait training and conventional gait training and conventional gait training alone in patients with chronic stroke (> 6 to 12 months) [789]. Combined robotic and conventional gait training produced greater improvements in gait function than did conventional gait training alone. Similarly, combined robotic and conventional gait training yielded a superior effect on gait function, balance, and activities of daily living than did conventional gait training in patients with subacute stroke (< 6 weeks to 6 months). These collective findings suggest that robot-assisted gait training provides an additive effect to conventional gait training in patients with hemiparetic stroke, regardless of the stroke phase condition.

Table 2 presents the summary of 11 randomized controlled trials that compared the effects of exoskeleton robot-assisted gait training with conventional gait training on motor recovery, gait function, and balance in hemiparetic stroke. Mayr et al. [10] and Husemann et al. [11] reported greater augmented effects from combined exoskeleton robot-assisted gait training combined with conventional gait training compared with conventional gait training alone in 10 subacute stroke patients. Hornby et al. [12] conducted a randomized controlled study that compared the effects of exoskeleton robot-assisted gait training and manual facilitation using an assist-as-needed paradigm on gait function in patients with chronic stroke. Moreover, Hidler et al. [13] also conducted a multicenter randomized trial that investigated the usefulness of Lokomat robot-assisted therapy in 72 patients with subacute stroke and found less effective in the 6-minute walk test (6MWT) and Functional Ambulation Category (FAC) tests than the conventional gait training. Similarly, Hornby et al. [12] found that Lokomat robot-assisted therapy in 62 patients with chronic stroke was not superior to the conventional gait training. Such inconsistent results in the Lokomat robot-assisted studies may result from different experimental design and testing methods utilized [10], lack of volitional neuromuscular control [14], restricted pelvic and trunk movement control [1516], arm swing, as well as altered acceleration and deceleration from pre-swing to initial contact [17].

The present review highlights current clinical evidence regarding exoskeleton and end-effector robot-assisted gait training approaches in subacute and chronic stroke patients. A recent Cochrane review [18] reported superior benefits with robot-assisted gait training in stroke rehabilitation. However, the types of robot-assisted gait training devices and their clinical effects are undetermined. Therefore, our review has focused on the types of robot-assisted gait training devices and their clinical effects in robotic assisted gait training. A meta-analysis of robot-assisted gait training studies is challenging owing to diverse robotic devices, heterogeneous subject characteristics, and varying experimental designs. Therefore, the present review attempted to provide clinical evidence as to which types of exoskeleton type or end-effector type devices are more effective for stroke rehabilitation and locomotor recovery and function. Taken together, the findings showed that combined robot-assisted gait training was advantageous in stroke rehabilitation, as it offers additive special therapeutic effects that were not afforded by conventional therapy alone. Most importantly, the robot-assisted gait training paradigm renders intensive, repetitive, accurate kinematic feedback and symmetrical gait practice, while reducing therapist labor, which is often not affordable in current stroke rehabilitation care [19]. Furthermore, the current robot-assisted gait training incorporates virtual or augmented reality to motivate patients, as well as to provide fun and ecologically valid gait training. Since the additive effect of robot-assisted gait training is well established, a prospective study is warranted to determine if robot-assisted gait training is superior to conventional stroke locomotor rehabilitation alone or robot-assisted gait training combined with conventional gait training. It would be of great interest to examine the effects of repetitive robot-assisted gait training on neuroplasticity and associated locomotor recovery in a stroke population.