Classifying Wheelchair Propulsion Strategies: A New Bottom-Up Approach Based on Kinematic Data Analysis

Wheelchair propulsion is closely linked to the physical condition of wheelchair users. Previous studies typically utilized a top-down approach in studying wheelchair propulsion strategies. Participants were pre-grouped based on injury level and each group was examined in terms of propulsion strategies. This study adopts a bottom-up classification using cluster analysis to uncover overlooked factors in wheelchair propulsion strategies. We recruited 29 participants with spinal cord injuries, and let them perform wheelchair propulsion on a wheelchair roller ergometer. After normalizing upper limb joint angle waveforms and conducting principal component analysis, we applied K-means cluster analysis and identified patterns in three clusters with the highest silhouette scores. The results reveal that propulsion strategies may be influenced not only by level of injury but also by various factors such as post-injury adaptation and exercise experience. Particularly, the cluster with higher physical activity level shows statistically higher speed values compared to other clusters. These findings highlight that appropriate training for residual muscle strength and individual wheelchair adaptation may play important roles in developing effective propulsion strategies.