Using wearable technology to inform clinical decision-making in upper-limb outpatient neurorehabilitation
Year: 2026
Type: dissertation
Abstract: Therapists in outpatient upper-limb (UL) neurorehabilitation for stroke and spinal cord injury patients lack objective data on home hand use, relying on noisy self-reports that hinder tailored therapy planning. This thesis develops a clinical decision support system (CDSS) using egocentric video to provide high-resolution, functional insights, with future work targeting video-centric querying, accessible wearables, and clinical trials to enhance recovery. In this thesis, we employ a user-centered, iterative approach to design and validate the CDSS. Qualitative interviews and observations with 10 therapists across a rehabilitation hospital network identified key challenges, including inadequate home performance data and time constraints for clinical assessments. Through focus groups and iterative co-design with seven clinicians, a prototype dashboard was developed, featuring time series graphs of hand interaction metrics (e.g., frequency and duration), activity breakdown charts, curated video snippets, and patient notes to contextualize functional performance. A pilot study with five patient-therapist dyads validated the system, revealing therapists' strong preference for video footage, which provided rich, actionable insights into patients' home environments and functional deficits compared to less interpretable graphs. However, implementation barriers, such as time demands and head-mounted camera usability, were noted. To address the impracticality of manual video review, an object-centric algorithm classified activities of daily living (e.g., self-feeding, meal preparation) from a dataset comprised of individuals with SCI performing unconstrained activities of daily living at home. This can enable activity-based video filtering, aligning with clinician workflow needs. No major differences emerged between stroke and spinal cord injury rehabilitation requirements, suggesting shared UL monitoring needs. Limitations include small sample sizes, hardware usability challenges for patients with severe impairments, and data imbalances affecting algorithm performance. Future work will address these through multi-site studies, user-friendly wearables (e.g., glasses-based cameras), co-designed clinical protocols, and randomized trials to quantify recovery impact using standardized measures. By establishing a scalable pipeline for integrating wearable technology into UL neurorehabilitation, this thesis enhances therapy personalization and lays the foundation for equitable, evidence-based clinical solutions.
Source: TSpace (University of Toronto)
Author: Kadambi, Adesh
Language: English
Subfield: Rehabilitation
Field: Medicine
Domain: Health Sciences
Open Access status: green
OpenAlex ID: W7204021060