Understanding Human Balance Strategies Through Clustering of Motion Analysis Data

Balance is the result of complex interactions among individuals, tasks, and the environment. It plays a crucial role in daily activities, relying on sensory systems and joint movement strategies to maintain postural stability. Effective balance not only improves movement and physical activities but also enhances overall quality of life. Balance strategies, such as ankle, hip, knee, and stepping strategies, are essential for maintaining stability in various situations and environments. This study focuses on utilizing public data from the Clinical Test of Sensory Interaction on Balance (CTSIB) to assess balance strategies. Traditional methods rely on subjective observations, limiting the identification of nuanced strategy combinations. To overcome this, motion capture (mocap) data and unsupervised machine learning techniques, clustering and dimensionality. reduction using Agglomerative, Gaussian Mixture, and K-Means and classify balance strategies. The results demonstrate the potential of this approach to identify distinct clusters of balance strategies, offering a more detailed, objective, and comprehensive understanding of postural stability. Among the clustering algorithms applied, K-Means proved to be particularly effective, achieving reliable groupings with consistent results. The Silhouette Scores indicated that k=3, identified as the optimal number of clusters from the elbow test, provided the best balance between cohesion and separation for K-Means. This suggests that K-Means delivers interpretable clusters of balance strategies. The three possible outcomes are good balance, A slight unbalance and a poor balance. This innovative methodology advances the field of balance assessment by providing a clear and systematic approach to classifying balance strategies, allowing for better understanding and targeted interventions.