Extended Reality Biofeedback for Functional Upper Limb Weakness: Mixed Methods Usability Evaluation

The perception-action cycle enables humans to adapt their behaviors by integrating sensory feedback into motor actions. Functional neurological disorder (FND) disrupts this cycle, leading to maladaptive motor responses and a diminished sense of agency. FND includes functional seizures, movement disorders, and cognitive impairments, significantly affecting quality of life. Recent advancements in extended reality (XR) neurotechnologies provide opportunities for novel rehabilitation approaches, leveraging visual and haptic feedback to retrain motor control and restore agency in individuals with functional limb weakness.

This study aimed to co-design and evaluate an XR-based biofeedback platform for upper-limb rehabilitation in FND, incorporating multisensory feedback (visual and haptic) to enhance motor retraining.

A mixed methods design was used. In phase 1, a Delphi survey (N=20, patients with FND) identified key user requirements, emphasizing customizability, real-time feedback, accessibility, and comfort. These insights guided the codevelopment of an XR biofeedback platform. In phase 2, a co-design workshop with 6 participants (3 FND patient representatives and 3 health care professionals) evaluated the usability of 3 XR training tasks: virtual reality (VR) relaxation task, a guided meditation in a VR calming environment; XR position feedback task ("Hoop Hustle"), a VR-based motion task requiring arm movements to interact with virtual objects, providing real-time positional biofeedback; and XR force feedback task, a haptic robot-assisted exercise using the Human Robotix System (HRX-1) haptic device, applying resistive forces to guide upper limb movements. Participants completed system usability scale (SUS) questionnaires and provided qualitative feedback, which was analyzed using NVivo (QSR International) thematic analysis.

The XR position feedback task achieved the highest usability ratings, with 4 out of 6 participants scoring it above 85, indicating "excellent" usability. The VR relaxation task received polarized scores: 2 participants rated it highly (90 and 87.5), while 3 scored it poorly (mid-40s), citing motion discomfort and disengagement. The XR force feedback task had mixed usability outcomes (SUS range: 27.5-95.0), with 1 participant with functional dystonia struggling significantly (SUS 27.5), while others rated it between 62.5 and 95.0. Qualitative feedback emphasized comfort (lighter headsets and better ergonomic design), immersion and content quality (clearer visuals and reduced distracting audio prompts), personalization (adjustable settings for speed, difficulty, and force resistance), and accessibility (cost concerns and home usability considerations). Overall, participants viewed the XR biofeedback platform as highly promising but in need of fine-tuning.

This study demonstrates the feasibility and usability of an XR neurotechnology platform for FND rehabilitation, with strong acceptance of XR position feedback, mixed reactions to VR relaxation, and individual-specific usability outcomes for the force feedback task. Findings underscore the need for personalization features and hardware refinement. Future work will focus on enhancing usability, improving accessibility, and evaluating effectiveness in larger clinical trials.