"It Depends": Re-Authoring Play Through Clinical Reasoning in Wearable AR Rehab Games
Binyan Xu, Wei Wu, Soonhyeon Kweon, Casper Harteveld, Leanne Chukoskie · 2026 · Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems (CHI '26) · doi:10.1145/3772318.3790457
Summary
This CHI 2026 paper examines how lightweight, glasses-form-factor augmented reality (AR) rehabilitation games can be translated from laboratory prototypes into everyday physical therapy (PT) practice. The authors argue that most AR rehab research has stalled in controlled studies, with limited clinical uptake, and that recent advances in spatial computing (e.g., Snap Spectacles) create a timely opportunity to embed rehabilitative play into routine clinical workflows. The study uses a two-phase mixed-methods design. First, the researchers systematically reviewed 132 Snap Spectacles AR applications across eight genres, narrowing to 21 movement-related games, then 9 PT-suitable candidates, and finally 4 games used in the main study (3-Dots-Pinch, Squishy Run, Ball Game, ActionBall). Second, they conducted playtesting sessions with 14 licensed U.S. physical therapists (10 in the main study), spanning orthopedics, sports medicine, neurology, pediatrics, and ICU care, with 1-37 years of experience. Sessions combined think-aloud gameplay with semi-structured interviews probing movement relevance, engagement/cognition, and clinical suitability. Transcripts were coded using a general inductive approach (Cohen's kappa = 0.68), clustered into six analytic domains, and organized via a modified WHO ICF framework. The paper positions therapists' frequent phrase "It depends" not as evasion but as a generative design principle rooted in clinical reasoning — the adaptive logic PTs use to tailor interventions under patient and contextual variability.
Key findings
The analysis surfaced three interpretive lenses describing how PTs re-author AR games. (1) Play is Co-Authored: therapists act as meta-designers, adjusting movement ranges, body-part logics, target placement, and multi-dimensional difficulty (size, color, number, spatial envelope, temporal patterns, rule-based overlays) rather than relying on speed-only progression, which PTs found "just punishment." (2) Play is Situated: game viability depends on condition-aware adaptation (frozen shoulder, stroke, Parkinson's, dysmetria), rehab stage (ICU/acute vs. outpatient vs. sports), and environmental setting (hygiene in hospitals, trip hazards at home). AR glasses were particularly valued for digitizing PT toolkits — virtualizing balls, cones, and resistance bands when equipment is scarce or sterility matters. (3) Play is Dual: AR mediates both physical recovery and psychological support via sustained repetition, therapeutic reinforcement, self-efficacy rebuilding through "small wins," graded exposure for kinesiophobia, and cognitive stimulation through dual-task layering. The paper synthesizes eight design implications including Progressive Postural Flexibility, Therapist-Editable Play Spaces, multi-dimensional Difficulty Scaling, Extending PT Toolkits, Psychological Mediation through Play, and Motivational Layering. A worked "PT-in-the-Loop AR Darts" case study demonstrates how one mechanic can graduate from seated large targets to balance-ball throws across populations.
Relevance
For accessibility and rehabilitation practitioners, this paper reframes adaptive technology design around clinician agency rather than auto-tuning algorithms. Its central contribution — that "it depends" is a design principle, not a research problem — pushes back against one-size-fits-all gamification and argues for therapist-editable parameters, transparent controls, and safe presets aligned with clinical workflows. The ICF-based framework offers a shared vocabulary that lets AR developers and clinicians align digital mechanics with body functions, activities, participation, and environmental factors, directly supporting WCAG-adjacent principles of personalization and user control in assistive contexts. Limitations include a U.S.-only PT sample skewed toward early- and late-career participants, reliance on a single platform (Snap Spectacles), and absence of patient participants — so claims are design-oriented rather than evidence of therapeutic effectiveness. Still, the work provides an unusually concrete bridge between HCI rehab-game research and the realities of everyday clinical practice.
Tags: augmented reality · rehabilitation · physical therapy · clinical reasoning · embodied interaction · serious games · gesture interaction · wearable technology · spatial computing · co-design
Standards referenced: ICF (WHO International Classification of Functioning, Disability and Health)