Digital biomarkers derived from wearable sensors in neurorehabilitation: a scoping review.

BACKGROUND: Wearable-derived digital biomarkers are increasingly being investigated in neurological rehabilitation to provide objective and continuous assessment of physical function beyond conventional clinic-based measures. However, the clinical applications and measurement characteristics of these biomarkers across neurological populations remain unclear. OBJECTIVE: To synthesize the current evidence regarding wearable-derived digital biomarkers used in neurological rehabilitation and to summarize their clinical applications, biomarker domains, and measurement characteristics across neurological conditions. METHODS: A scoping review was conducted following the PRISMA-ScR framework. Electronic databases including PubMed/MEDLINE, ScienceDirect, Cochrane Library, and Google Scholar were searched for English-language studies published between January 2015 and May 2026. Eligible studies investigated wearable-derived digital biomarkers in neurological rehabilitation populations. RESULTS: A total of 56 studies were included in the review, with stroke (n = 29) and Parkinson disease (n = 10) representing the most frequently investigated neurological conditions. Inertial Measurement Unit (IMU) based systems were reported in 34 studies, making them the predominant wearable technology. Kinematic biomarkers were the most commonly investigated biomarker domain, particularly in stroke (16/29) and Parkinson disease studies (9/10). In contrast, activity-based (2/5) and physiological biomarkers (2/5) were more prevalent in spinal cord injury research. Although several studies reported favorable validity and reliability of wearable-derived measures, substantial heterogeneity was observed across studies in sensor configurations, biomarker definitions, and validation methodologies. CONCLUSION: Wearable-derived digital biomarkers demonstrate considerable potential for objective and ecologically valid monitoring in neurological rehabilitation. However, methodological heterogeneity and limited longitudinal evidence indicate that clinical implementation remains at an early stage of development.