The purpose of this study was to explore the ability of Kohonen neural network self-organizing maps (SOM) to visualize and characterize different movement patterns during sidestepping. The marker trajectories of 631 sidestepping trials were used to train a SOM. Out of 63731 input vectors, the SOM identified 1250 unique stick figures, determined by the markers. Visualizing the movement trajectories and adding the latent parameter time, allows for the investigation of different movement patterns. Additionally, the SOM can be used to identify zones with increased injury risk, by adding more latent parameters which opens the option to monitor athletes and give feedback. The results highlight the ability of unsupervised learning to visualize movement patterns and to give further insight into an individual athlete`s status without the necessity of a-priory assumptions.
© Copyright 2022 ISBS Proceedings Archive (Michigan). Northern Michigan University. Published by International Society of Biomechanics in Sports. All rights reserved.