Smartphone-based telerehabilitation for post-stroke motor recovery assessment

Thesis Abstract

This study addresses the growing need for accessible, objective, and scalable assessment of motor recovery after stroke through smartphone-based telerehabilitation. Despite advances in digital health, there remains a gap in reliable remote measurement of upper-limb motor function that can be feasibly integrated into routine post-stroke rehabilitation. The aim is to evaluate the validity, reliability, and clinical utility of a smartphone-based telerehabilitation platform for assessing post-stroke motor recovery, and to explore its impact on patient adherence and functional outcomes. Specific objectives include (1) to determine concurrent validity of smartphone-based motor metrics against gold-standard clinical scales (Fugl-Meyer Assessment for Upper Extremity, activity of daily living measures) and kinematic measures captured via inertial sensors; (2) to assess test-retest and inter-device reliability across multiple smartphone models and operating systems; (3) to examine the platform’s sensitivity to change over a 12-week intervention period; (4) to identify determinants of user engagement and adherence using the Unified Theory of Acceptance and Use of Technology (UTAUT) and the Technology Acceptance Model; and (5) to model the relationship between remote motor assessments and real-world functional performance via regression analysis, controlling for age, baseline impairment, and time since stroke. The methodology employs a mixed-methods, explanatory sequential design. In the quantitative strand, a multi-center cohort of 240 adult stroke survivors (mean age 62.5 ± 9.8 years; time since stroke 3–18 months) will be recruited from five neurorehabilitation centers. Participants will complete a 12-week remotely supervised program using the smartphone platform, with weekly motor tasks and continuous passive monitoring. Primary data include smartphone-derived motor metrics (range of motion, grip strength proxies, movement smoothness, speed, and coordination indices) and standardized clinical assessments (Fugl-Meyer Upper Extremity, Box and Blocks Test, and the Barthel Index) at baseline, week 6, and week 12. Secondary data include adherence indicators (login frequency, task completion rate) and passive data (daily activity, sleep quality). Data collection instruments comprise validated digital sensors within the app, Bluetooth-enabled inertial measurement units when available, and clinician-rated scales administered via teleconsultation. Reliability will be examined through intraclass correlation coefficients (ICCs) for repeated measures and Bland-Altman analyses across devices; validity will be evaluated through Pearson correlations and concordance with clinical scales. Sensitivity to change will be assessed using effect sizes (Cohen’s d) and standardized response means. Qualitative data will be gathered from a purposive subsample of 40 participants and 15 clinicians through semi-structured interviews to explore perceived usefulness, ease of use, and barriers/facilitators of adoption, analyzed via thematic analysis aligned with the Consolidated Framework for Implementation Research (CFIR). The integration of quantitative and qualitative findings will facilitate a convergent synthesis to explain observed outcomes. Data analysis for the quantitative component will utilize multilevel mixed-effects models to account for repeated measures and site variability, with fixed effects for time, baseline impairment, age, and interaction terms to assess differential trajectories. Regression analyses will determine the extent to which remote motor metrics predict real-world functional gains, with model assumptions checked and collinearity diagnostics performed. A subgroup analysis will examine differential effects by baseline impairment severity and time since stroke. Thematic analysis will be conducted using NVivo, following a six-step coding framework to derive themes related to usability, perceived impact on daily activities, and integration into clinical workflows. Expected findings include strong convergent validity between smartphone-based metrics and established clinical measures, high test-retest reliability across devices, and meaningful sensitivity to change over 12 weeks. It is anticipated that higher engagement with the platform will correlate with greater functional improvements, and that remote assessments will moderately predict real-world performance after controlling for confounders. The study is expected to contribute to knowledge by validating a scalable, ecologically valid method for remote motor assessment post-stroke, elucidating factors that influence adoption, and informing guidelines for integrating smartphone-based telerehabilitation into standard care. The main conclusion will emphasize the platform’s potential to augment clinical decision-making, monitor progress in real time, and support timely adjustments to therapy intensity. Recommendations will address optimization of user interfaces, integration with electronic health records, data governance, and strategies to enhance adherence among older adults and those with severe paresis.