individual.s degree of ankle pronation during running is often considered an identifying characteristic of their running style. Yet there is no consensus for defining different types of pronation in order to distinguish among runners [1]. Similar disagreement prevails on the thresholds that separate forefoot and rearfoot strike patterns. The objective of this study was to examine how well different methods of pronation and footstrike assessment could classify runners into distinct groups using support vector machines. Kinematic data was collected from 50 recreational runners during overground and treadmill running trials. After collection, each subjects. kinematic data were assessed as one of four types of frontal plane motion (Supinator, Neutral, Pronator, or Over-pronator) and one of three types of sagittal plane motion (Forefoot, Midfoot, or Rearfoot striker). Three assessment methods were utilized: Kinematic, based on motion capture data, Video, based on visual analysis, and Self, which came from the subjects themselves. A support vector machine (SVM) was trained to distinguish between normalized kinematic trajectories using each assessment method to define pronation styles. Binomial tests determined if classification by the SVM was significant. The results show Kinematic and Video assessment methods reveal significant differences between all types of characteristic foot motion in both the frontal (Table 1) and sagittal (Table 2) planes. However self-assessments did not produce significantly separable groups of runners. That both Kinematic and Video methods could separate groups of runners reliably suggests that the definitions on which they were based represent meaningfully different running patterns. Conversely, the inability of self-assessment to separate groups indicates that this information does not reflect independent types of runners. Vector based methods such as SVM may prove useful in standardizing definitions of pronation and footstrike running styles.
© Copyright 2014 International Calgary Running Symposium, August 14-17, 2014. All rights reserved.