Reliability and criterion validity of a wearable IMU for ankle joint position sense: agreement with isokinetic dynamometry and discrimination of chronic ankle instability

Abstract

Abstract Background Chronic ankle instability (CAI) impairs joint position sense (JPS) and increase the risk of reinjury. Inertial measurement units (IMUs) are cost-effective and portable, but their reliability and validity for ankle JPS assessment remain uncertain. This study aimed to evaluate the reliability and concurrent validity of IMU-based ankle JPS assessment against an isokinetic dynamometer and to compare JPS errors between individuals with CAI and healthy controls. Methods In a prospective, assessor-blinded, cross-sectional study, adults with CAI (Cumberland Ankle Instability Tool ≤ 24) and matched controls performed active joint position reproduction tasks in plantarflexion, dorsiflexion, inversion, and eversion with vision occluded. Two IMU sensors recorded joint angles concurrently with a Contrex isokinetic dynamometer (reference standard). Absolute error (AE, °) was the primary outcome. Reliability was quantified using intraclass correlation coefficients (ICC) of intra-session reliability, standard error of measurement, and minimal detectable change; concurrent validity was examined with Bland–Altman analyses. Between-group differences and diagnostic utility were evaluated using group comparisons and receiver operating characteristic (ROC) analyses. Results Both the isokinetic dynamometer and the IMU demonstrated good-to-excellent intra-session reliability (dynamometer ICC 0.91–0.96; IMU 0.83–0.90), with larger SEM and MDC₉₅ values for the IMU. Bland–Altman analyses showed small negative biases (approximately − 0.4° to − 2.6°) and 95% limits of agreement generally within about ± 5° for plantarflexion, dorsiflexion, and inversion, with somewhat wider limits in eversion. Compared with controls, participants with chronic ankle instability exhibited significantly greater absolute JPS errors in plantarflexion, dorsiflexion, and inversion on both devices, and in eversion when assessed by the IMU. ROC analyses showed excellent discrimination for plantarflexion (AUC 0.93 for the dynamometer, 0.91 for the IMU), good discrimination for dorsiflexion and inversion, and limited, non-significant discrimination for eversion. Conclusion IMU-based ankle JPS testing demonstrates good reliability and clinically acceptable agreement with isokinetic dynamometry while detecting proprioceptive deficits in CAI, particularly in plantarflexion, dorsiflexion, and inversion. It appears to be a practical tool for clinical and sports settings, although eversion measurements were more variable and future studies should confirm these findings in larger samples and dynamic, multi-planar tasks.