Key result
Multimodal XGBoost framework classifies balance dysfunction in older adults with ~84% accuracy.
Why the study?
Conventional subjective scales and single-modality measures may not adequately capture the integrated motor, cortical, and cognitive processes underlying postural control in older adults.
Can a multimodal assessment framework using COP, sEMG, and fNIRS accurately classify balance dysfunction and identify functional patterns in older adults?
Cross-Sectional (n=81)
Can a multimodal assessment framework using COP, sEMG, and fNIRS accurately classify balance dysfunction and identify functional patterns in older adults?
A multimodal assessment combining biomechanical, muscular, and cortical data can accurately classify balance dysfunction and identify heterogeneous functional patterns in older adults.
Answers are grounded in this paper and cite their sources.
Supports multimodal profiling of balance dysfunction; hypothesis-generating and requires prospective validation before clinical use.
BACKGROUND: Balance dysfunction in older adults is a major risk factor for falls. Conventional subjective scales and single-modality measures may not adequately capture the integrated motor, cortical, and cognitive processes underlying postural control. Using a multimodal feature set, this study aimed to develop an interpretable framework for dysfunction-related classification and for exploring heterogeneity in balance-related functional patterns. METHODS: A total of 81 community-dwelling older adults (≥ 60 years old) were recruited and divided into balance dysfunction group (n = 38, BBS ≤ 45) and healthy control group (n = 43, BBS > 45) according to the Berg Balance Scale (BBS). During six stance tasks, center of pressure (COP) trajectories, lower-limb surface electromyography (sEMG), and cortical activation measured by functional near-infrared spectroscopy (fNIRS) were synchronously recorded. Group × task effects were tested using two-way repeated-measures ANOVA. Multimodal features were then used for supervised classification (XGBoost) and for clustering-based exploration of heterogeneity within dysfunction-related multimodal profiles using K-means clustering. RESULTS: Under more challenging postural conditions, the dysfunction group showed larger anteroposterior COP oscillation range, velocity, and area. Muscle activation patterns showed lower rectus femoris contribution together with higher biceps femoris and gastrocnemius activation under selected task conditions, accompanied by altered muscle synergy organization. fNIRS revealed greater activation in premotor cortex, supplementary motor area, and prefrontal cortex regions in the dysfunction group. In classification, XGBoost achieved the best overall performance among the tested models, with 83.56% accuracy. Clustering analysis further identified three functional patterns with graded differences in BBS and Montreal Cognitive Assessment (MoCA) scores, suggesting heterogeneity in dysfunction-related multimodal profiles. CONCLUSION: Balance dysfunction in older adults is associated with a neuro-muscular-cognitive profile involving impaired postural control, altered lower-limb muscle recruitment, and increased cortical activation under more demanding task conditions. The proposed multimodal framework provides an interpretable approach for classification and for exploring heterogeneity in dysfunction-related functional patterns. Further validation in larger cohorts and simplified sensor configurations will be needed to support broader practical application.
No takes yet. Share an insight, caveat, or question.
Sun et al. (2026) conducted a cross-sectional in Balance dysfunction (n=81). Balance dysfunction vs. Healthy controls was evaluated on Classification accuracy of multimodal features using XGBoost. A multimodal framework using XGBoost classified balance dysfunction in older adults with 83.56% accuracy, identifying distinct neuro-muscular-cognitive profiles.
Synapse has enriched 10 related papers. Consider them for comparative context: