An exploratory study evaluating the effectiveness of a data driven approach to identifying coordinative features that are associated with sprint velocity
KeywordBiomechanics, machine learning, coordination, sprint performance, objective movement assessment
MetadataShow full item record
AbstractSprint performance is multifactorial in nature and is dependent on a variety of coordination and motor control features. During the sequential phases of a sprint, the athlete completes a series of spatiotemporal coordination strategies to achieve the fastest possible velocity. The overall aim of the study was to leverage wearable sensor technology and data- driven tools to objectively assess the kinematic and neuromuscular determinants of optimal sprint velocity from a large dataset of university-aged sprinters. To achieve this, we recruited participants to run three 60 m sprints as fast as possible, while being outfitted with wireless electromyography (EMG) and a full-body inertial measurement unit (IMU) suit to obtain full- body 3D kinematics. Five strides about peak sprint velocity were selected and used for inputs into a principal components analysis (PCA). Significant stepwise multivariable regression models were generated for both kinematic and EMG features identified using PCA, with the kinematic model outperforming the EMG model as the kinematic model displayed a higher R2 value. This suggests that the kinematic dataset used in this study is a better predictor of sprint performance when compared to the EMG dataset, and that both may be viable options in the development of data-driven objective sprint coaching tools.
The following license files are associated with this item:
- Creative Commons