Background:

Robotics has emerged as a promising avenue for gait retraining of persons with chronic hemiparetic gait and footdrop, yet there is a gap regarding the biomechanical adaptations that occur with locomotor learning. We developed an ankle exoskeleton (AMBLE) enabling dorsiflexion assist-as-needed across gait cycle sub-events to train and study the biomechanics of motor learning stroke. This single-armed, non-controlled study investigates effects of nine hours (9 weeks × 2 sessions/week) locomotor task-specific ankle robotics training on gait biomechanics and functional mobility in persons with chronic hemiparetic gait and foot drop. Subjects include N = 16 participants (8 male, 8 female) age 53 ± 12 years with mean 11 ± 8 years since stroke. All baseline and post-training outcomes including optical motion capture for 3-D gait biomechanics are conducted during unassisted (no robot) over-ground walking conditions.

Findings:

Robotics training with AMBLE produced significant kinematic improvements in ankle peak dorsiflexion angular velocity (°/s, + 44 [49%], p < 0.05), heel-first foot strikes (%steps, + 14 [15%], p < 0.01) toe-off angle (°, + 83[162%], p < 0.05), and paretic knee flexion (°, + 20 [30%], p < 0.05). Improvements in gait temporal-spatial parameters include increased paretic step length (cm, + 12 [20%], p< 0.05), reduced paretic swing duration (%GC, -3[6%], p < 0.05), and trend toward improved step length symmetry (-16 [11%], p = 0.08). Functional improvements include 10-meter comfortable (m/s, + 13 [16%], p < 0.01) and fastest (m/s, + 13 [15%], p<0.01) walking velocities, 6-minute timed walk distance (m, + 16 [19%], p < 0.01) and Dynamic Gait Index scores (+15 [15%], p < 0.01). Subjects' perceived improvements surpassed the minimal clinically important difference on the Stroke Impact Scale (SIS) mobility subscale (+11 [19%], p < 0.05).

Conclusions:

AMBLE training improves paretic ankle neuromotor control, paretic knee flexion, and gait temporal-distance parameters during unassisted over-ground walking in persons with chronic stroke and foot drop. This locomotor learning indexed by an increase in volitional autonomous (non-robotic) control of paretic ankle across training translated to improvements in functional mobility outcomes. Larger randomized clinical trials are needed to investigate the effectiveness of task-specific ankle robotics, and precise training characteristics to durably improve gait, balance, and home and community-based functional mobility for persons with hemiparetic gait and foot drop. Clinical trial identifier NCT04594837.