# Adaptive control ankle robotics training durably improves gait biomechanics in chronic hemiparetic stroke and footdrop

## TL;DR

A wearable ankle robot improved walking for stroke survivors with foot drop, with benefits lasting two months after training.

## Contribution

A wearable ankle exoskeleton with adaptive control improved gait biomechanics and functional mobility in chronic stroke patients.

## Key findings

- AMBLE training improved gait biomechanics like toe clearance and ankle dorsiflexion angular velocity.
- Functional outcomes such as walking speed and 6-minute walk distance improved after training.
- Most biomechanical improvements were retained two months after training ended.

## Abstract

Robotics has emerged as a promising avenue for gait rehabilitation after stroke. We developed a wearable ankle exoskeleton (AMBLE) for dorsiflexion assist-as-needed training with adaptive control timing to individualize assistance across gait cycle sub-events. This single-armed, non-controlled study investigates effects of 9 weeks x 2 sessions/week robotics training on walking function in persons with chronic stroke and foot drop, and durability 2 months after training ends.

Subjects included N = 24 participants (12 male, 12 female) age 57 ± 13 years with mean 10 ± 9 years since stroke. All baseline and post-training outcomes included optical motion capture for 3-D gait biomechanics and were conducted during unassisted (no robot) over-ground walking conditions. AMBLE training improved select gait biomechanics outcomes including maximum toe clearance (mm, pre- 69±28 versus post- 79±30, p < 0.01), ankle peak dorsiflexion angular velocity (°/s, 35±32 versus 47±40, p < 0.01), heel-first foot strikes (%steps, 31±41 versus 44±43, p < 0.01), and paretic step length (cm, 37±16 versus 40±14, p < 0.01. Functional outcomes that improved with training included 10-meter self-selected (m/s, 0.66±0.24 versus 0.70±0.23, p < 0.01) and fastest comfortable velocities (m/s, 0.80±0.31 versus 0.86±0.30, p < 0.01), Dynamic Gait Index (points, 14±5 versus 17±3, p < 0.01), 6-minute walk distance (m, 252±106 versus 280±109, p < 0.01), and Stroke Impact Scale-Mobility (0-100, 280±109 versus 88±10, p < 0.01); all achieving minimal clinically important differences, except walking velocities. Durability testing 2 months after cessation of robotics training showed retention of most biomechanical improvements, including maximum toe clearance (mm, 78.4±26.1, p < 0.01), dorsiflexion angular velocity (°/s, 42.5±37.5, p < 0.01), and heel-first foot strikes (%steps, 46±43, p < 0.01), and most functional outcomes. Notably, durability testing revealed emergence of two new kinematic improvements: increased knee flexion (deg, 33.1±16.5 versus 36.9±17.7, p < 0.01) and hip flexion (deg, 36.8±9.5 versus 39.6±9.2, p < 0.05), while hip abduction and hip hike do not change.

Nine hours of AMBLE ankle robotics training across 9 weeks durably improves gait biomechanics and functional mobility for persons with chronic stroke and foot drop, even decades post-stroke. Further studies are needed to investigate precision adaptive control robotics for stroke and other mobility disability conditions.

Trial registration: Clinical trial identifier: NCT04594837.