Can digital gait biomarkers complement the WHO intrinsic capacity framework in fall risk stratification? A cross-sectional exploratory analysis

The World Health Organization (WHO) Intrinsic Capacity (IC) framework serves as a holistic construct for healthy aging and fall risk screening. However, current IC assessments predominantly rely on subjective reports or macroscopic physical tests, which may lack the sensitivity to detect subtle kinematic deficits. This study aimed to determine whether integrating objective, wearable-derived digital gait biomarkers into the WHO IC framework provides measurable incremental value for identifying older adults with a history of falls. We conducted a cross-sectional analysis of 843 community-dwelling older adults in Beijing, China. Fall history and IC domains were assessed using standard protocols. Gait performance was quantified via a wearable system during a 12-m walk, yielding 45 spatiotemporal and kinetic features. LASSO regression was employed to handle multicollinearity and identify the most discriminative gait markers to construct a composite Gait Score. The incremental discriminative ability of adding the Gait Score to the standard IC model was evaluated using the Area Under the Curve (AUC), Net Reclassification Improvement (NRI), and Integrated Discrimination Improvement (IDI). Among participants, 105 (12.5%) reported falls. LASSO regression identified five gait variables included in the Gait Score: step length, landing control force, toe-off angle, step-duration asymmetry, and initial support time. In fully adjusted multivariable models, both IC impairment (OR = 1.330, 95% CI 1.082–1.634) and the composite Gait Score (OR = 2.583, 95% CI 1.123–5.945) were independently associated with fall status. Adding the Gait Score to the IC-based model yielded a statistically detectable but modest improvement in discrimination (IDI = 0.007, P = 0.044). In a sensitivity analysis that repeated feature selection and score derivation within the bootstrap procedure, the AUC for Model 3 decreased from 0.676 (apparent) to 0.607 (optimism-corrected), underscoring the exploratory nature of the incremental finding. In this cross-sectional sample, wearable-derived gait characteristics were associated with past-year fall status and provided limited, exploratory complementary discrimination beyond the IC impairment score. These findings require prospective external validation before the approach can be considered for clinical risk stratification.