Multimodal assessment of balance dysfunction in older adults: from classification to clustering-based functional...

Area of Science:

• Gerontology and Biomechanics
• Neuroscience and Motor Control
• Medical Technology and Data Analysis

Background:

• Balance dysfunction in older adults is a significant fall risk factor.
• Conventional assessment methods may not fully capture complex postural control mechanisms.
• Need for integrated motor, cortical, and cognitive assessments.

Purpose of the Study:

• To develop an interpretable multimodal framework for classifying balance dysfunction in older adults.
• To explore the heterogeneity of balance-related functional patterns.
• To integrate center of pressure (COP), electromyography (sEMG), and functional near-infrared spectroscopy (fNIRS) data.

Main Methods:

• Recruited 81 older adults (≥60 years), divided into balance dysfunction (n=38) and healthy control (n=43) groups based on Berg Balance Scale (BBS).
• Synchronously recorded COP, lower-limb sEMG, and fNIRS during six stance tasks.
• Employed XGBoost for classification and K-means clustering for heterogeneity analysis.

Main Results:

• The dysfunction group exhibited greater COP oscillations, altered muscle synergies (e.g., reduced rectus femoris, increased biceps femoris/gastrocnemius), and heightened cortical activation (premotor, supplementary motor, prefrontal cortex).
• XGBoost classification achieved 83.56% accuracy.
• Clustering identified three distinct functional patterns with varying BBS and Montreal Cognitive Assessment (MoCA) scores.

Conclusions:

• Balance dysfunction is linked to a neuro-muscular-cognitive profile characterized by impaired postural control, altered muscle recruitment, and increased cortical activity.
• The multimodal framework offers an interpretable method for classification and understanding functional heterogeneity.
• Further validation in larger cohorts and simplified systems is recommended for practical application.