Wearable-Derived Digital Biomarkers for Mobility Assessment and Risk Stratification in Older Adults: A Scoping Review

The global population aged ≥60 years will reach 2.1 billion by 2050. Mobility decline is a critical precursor of disability, hospitalization, and mortality in older adults, yet clinic-based assessments may have limited ecological validity and miss subtle deteriorations observable only under free-living conditions. Wearable technology offers continuous objective monitoring through digital biomarkers that may enable earlier intervention.

To systematically map current evidence on digital biomarkers derived from wearable technology for mobility assessment and risk stratification in older adults (≥60 years), examining biomarker types, sensor devices, validation approaches, clinical applications, and evidence gaps. Evidence was stratified descriptively across cross-sectional mobility impairment, fall/frailty risk stratification, and longitudinal/prognostic decline.

Following Joanna Briggs Institute (JBI) methodology and PRISMA-ScR guidelines, we searched four databases (Scopus, PubMed/MEDLINE, Web of Science Core Collection, and IEEE Xplore) from 1 January 2019 to 31 March 2026. The protocol was prospectively registered (OSF; https://doi.org/10.17605/OSF.IO/PABFT). Two independent reviewers conducted screening and data extraction; disagreements were resolved by a third reviewer. A narrative synthesis was employed. Formal risk-of-bias assessment was not undertaken, consistent with JBI scoping-review methodology.

From 3,345 identified records, 97 primary empirical studies met inclusion criteria after removing 2,796 duplicates and applying predefined eligibility (Cohen's κ = 0.86 title/abstract; κ = 0.88 full-text). Seven biomarker categories were identified (non-mutually exclusive; percentages sum to >100%): gait spatiotemporal parameters (38.1%), daily-life mobility metrics (24.7%), frailty-related biomarkers (18.6%), cognitive-physical dual biomarkers (15.5%), circadian rhythm parameters (12.4%), gait dynamics/complexity (9.3%), and consumer device metrics (8.2%). Validation performance, where comparable, ranged from ICC 0.70-0.98 (reliability vs reference systems) and AUC 0.73-0.94 for fall/frailty discrimination; these are study-specific, not pooled estimates. Among circadian biomarkers, one large registry study (UK Biobank age-stratified analysis) reported a CosinorAge hazard ratio of 1.58 per year for all-cause mortality (95% CI 1.52-1.64), warranting prospective validation in dedicated older-adult cohorts.

Wearable-derived digital biomarkers show promising, study-specific criterion and discriminative validity for mobility assessment and fall/frailty risk stratification in older adults across multiple clinical domains; however, direct longitudinal evidence for detecting mobility decline remains limited and heterogeneous. Evidence consistency and maturity vary across biomarker domains and are descriptive rather than derived from a formal risk-of-bias assessment. Critical gaps remain in longitudinal validation, standardised protocols, implementation science, and equity-focused research.