Hybrid Brain/Neural Exoskeleton Enables Bimanual ADL Training in Routine Stroke Rehabilitation.

Abstract

Severe upper limb motor impairment is one of the most disabling consequences of stroke. Although brain-controlled rehabilitation technologies, such as brain/neural exoskeletons (B/NE), have been shown to be effective in promoting motor recovery, their clinical adoption remains limited because of insufficient integration of B/NE into existing clinical workflows. Here, we introduce and validate a fully portable B/NE system that overcomes this limitation by relying on brain (electroencephalography) and ocular (electrooculography) signals to restore bimanual activities of daily living within a novel therapeutic framework. In this pilot study, we tested the feasibility of the novel approach in 5 stroke survivors (mean age, 51 years; SD=14.8) undergoing inpatient neurorehabilitation. Stroke survivors aged 18 to 80 years, who exhibited hemiparesis and sufficient cognitive ability to understand and follow instructions, were invited to participate in a 1-hour training session. This session included system setup and calibration, followed by performing B/NE-supported, self-paced bimanual activities of daily living. As primary outcome measures, we assessed control accuracy, the ability to reliably modulate electroencephalography and electrooculography signals, and time to initialize, defined as the time required to react to cues and initiate the task, serving as a measure of control intuitiveness. In addition, participants' B/NE control performance during assisted training of bimanual activities of daily living, as well as setup preparation time, were assessed via direct observation. Participants demonstrated reliable control accuracy in using both brain (mean, 83%; SD=15.36) and ocular (mean=100%) signals, as well as intuitive control (time to initialize <2 s). All participants reliably controlled the B/NE performing a battery of 10 bimanual activities of daily living. Moreover, setup and calibration times remained below 20 minutes. These findings highlight the compatibility of the novel B/NE with existing clinical workflows and its feasibility to enable B/NE-supported stroke neurorehabilitation by facilitating seamless integration into clinical practice.