Human-centered Design of a Privacy-preserving Mixed-reality Exercise Program for Stroke
Gina Gwiazda1, Alison Scheid2, Edward Kim3, Yuri Cho3, Jeremy Fischer3, Siqi Li4, Mahmoud Elfar4, Erik Nelson3, Seonghyun Park3, Rachel Zeng3, Kamesh Krishnamurthy1, Jessica Mok2, Yasser Shoukry4, Bjoern Hartmann3, Sanjit A. Seshia3, Cathra Halabi1
1UCSF Department of Neurology, 2UCSF Department of Physical Therapy and Rehabilitation Science, 3UC Berkeley Department of Electrical Engineering and Computer Sciences, 4UC Irvine Department of Electrical Engineering and Computer Science
Objective:
Develop a proof-of-concept, privacy-preserving, mixed-reality exercise program for stroke rehabilitation using a multidisciplinary and stakeholder-driven approach.
Background:
Post-stroke rehabilitation is essential for functional recovery, but many patients have limited access to care. Telerehabilitation (TR) connects patients and providers remotely, and mixed-reality (MR) integrates virtual objects into real-world environments via headsets. TR and MR can make rehabilitation broadly accessible. This study’s interdisciplinary team designed a novel MR exercise program incorporating feedback from people with stroke-related lived experience.
Design/Methods:
The exercise program was designed and tested in 3 phases: 1) cross-domain knowledge sharing, 2) program development, and 3) workshop sessions. During Phase 1, clinicians (neurologists, therapists) and computer science researchers met weekly to exchange expertise regarding rehabilitation, cyber-physical systems, and a team-developed MR program (ScenicMR). Phase 2 comprised exercise protocol development and integration into ScenicMR (ScenicMR-Recovery and Rehabilitation, or ScenicRnR). Phase 3 comprised consented participants recruited into two study cohorts: 1) individuals with ongoing stroke-related impairments (“Patient”), and 2) individuals recovered from stroke or healthy volunteers related to individuals recovering from stroke (“Recovered/HV”). Both cohorts tested and evaluated ScenicRnR.
Results:
Phases 1 and 2 yielded MR-based exercises with physical and virtual objects, targeting domains including upper extremity activity and balance. Multi-layer privacy-preserving algorithms were developed to identify sensitive environmental information with options for user-directed privacy shielding. In Phase 3, “Patient” (n=4) characteristics ranges were age 49-66 years, 50% female, NIH Stroke Scale 4-5, modified Rankin Scale 2 (all), chronicity 15-247 months. “Recovered/HV” (n=4) characteristics ranges were age 50-70 years, 25% female. All participants expressed interest in TR with ScenicRnR and recommended gamifying exercises. Most (n=6 strongly agreed, 2 no response) anticipated comfort with caregiver-supported home use, though differing perspectives regarding role of caregiver involvement emerged.
Conclusions:
A human-centered design approach yielded a prototype MR stroke exercise program. Determining feasibility of remote deployment for TR is underway.
Disclaimer: Abstracts were not reviewed by Neurology® and do not reflect the views of Neurology® editors or staff.