Kinematic analysis of one-leg standing for locomotive syndrome screening using marker-less motion capture and machine learning: A cross-sectional study

• Marker-less kinematic analysis identifies early locomotive decline in older adults. • One-leg standing kinematics enhance accuracy of locomotive syndrome classification. • Machine learning improves LS stage detection, achieving AUC up to 0.904. • Approaches feasible in space-limited clinical and community settings. Locomotive syndrome (LS), caused by musculoskeletal dysfunction, is a major concern for older adults, contributing to physical decline, increased caregiving needs, and loss of autonomy. This study aimed to develop a novel LS screening method by analyzing kinematic data from the one-leg standing (OLS) test using a marker-less motion capture system combined with machine learning. A total of 294 community-dwelling older adults participated, and LS severity was evaluated using the LS risk test. Kinematic data were collected during the OLS test by marker-less motion capture system and analyzed with neural network models to classify LS severity into stage 1 or higher and stage 2 or higher. The proposed model demonstrated superior classification accuracy compared to conventional OLS time-based assessments, with area under the curve (AUC) values of 0.901 for stage 1 or higher and 0.904 for stage 2 or higher, exceeding the performance reported in previous studies. This study demonstrates the feasibility of integrating kinematic analysis with machine learning to screen LS. The developed model offers a noninvasive and efficient tool for LS detection, suitable for both telemedicine and community-based applications. By facilitating early detection of LS, this approach has the potential to improve enhance clinical outcomes and reduce caregiving burdens.