Inertial Sensor Reliability and Validity Across a Five-Level Surface Instability Gradation During Single-Leg Standing

Wearable inertial sensors offer a portable alternative to laboratory-grade force plates for postural stability assessment; however, their validity across progressively challenging balance tasks remains under-explored. This study evaluated the reliability and concurrent validity of inertially sensed metrics compared with force-plate-derived postural sway metrics across a five-level spectrum of unstable surfaces (Floor, Foam Pad, Rotating Disc, Air Disc, Bosu). Twenty-five healthy young women (22.1 ± 3.6 years, 1.64 ± 0.04 m, 58.44 ± 8.21 kg) performed five trials of single-leg standing (40 s each) on each surface. Postural sway was computed from antero-posterior (AP) and medio-lateral (ML) center of pressure (CoP) recordings using a force plate (Kistler, 9286 AA, Winterthur, Switzerland, sampling at 500 Hz) in synchronization with a lateral shank-mounted inertial sensor (Bionomadix BN-ACCL3, Biopac Systems, Inc., Santa Barbara, CA, USA, sampling at 100 Hz). In addition to reliability, a two-tiered analysis evaluated global concordance (unstandardized slopes) and method agreement (standardized z-scores). Intraclass correlation coefficients (ICCs) for the inertial sensor were excellent (range: 0.95-0.96), surpassing the force plate (range: 0.85-0.92) as trials accumulated. Analysis revealed moderate-to-good global concordance in the AP direction (r = 0.60, p = 0.001) and good-to-excellent in the ML one (r = 0.85, p < 0.001), validating the progressive intensifying effect of the surface graduation. Individual ranking agreement-evaluated via standardized z-scores-was also significant in both the AP (r = 0.61, p < 0.001) and the ML (r = 0.85, p < 0.001) directions, indicating a convergence into how the two modalities rank individual performance. Bland-Altman plots confirmed high absolute agreement between standardized scores, though a predictable proportional bias was observed in raw units, where the inertial sensor's underestimation of sway magnitude increased linearly with task difficulty. The five-level postural challenge graduation is a highly reliable framework for balance assessment. While the shank-mounted sensor exhibits proportional underestimation of sway magnitude compared to the CoP at extreme intensities, its high internal stability and sensitivity to task difficulty make it a valid and robust tool for longitudinal clinical monitoring.