Classifying Elite From Novice Athletes Using Simulated Wearable Sensor Data.

Area of Science:

• Biomechanics
• Sports Science
• Machine Learning

Background:

• Traditional movement screens rely on subjective visual assessment, leading to poor reliability.
• Objective movement analysis using machine learning can classify athletes by skill level.
• Optical motion capture is accurate but costly and time-consuming, limiting practical application.

Purpose of the Study:

• To determine if machine learning can classify elite and novice athletes using data from inertial measurement units (IMUs).
• To optimize the machine learning model architecture for improved classification accuracy.
• To develop an objective, cost-effective, and field-deployable movement assessment tool.

Main Methods:

• Analysis of motion capture data from 542 athletes performing seven dynamic screening movements.
• Application of principal component analysis (PCA) for pattern recognition.
• Utilizing machine learning algorithms with segment linear accelerations and angular velocities from IMUs as input for classification.

Main Results:

• Machine learning models achieved 75.1-84.7% accuracy in classifying athletes as elite or novice, depending on the movement.
• Classification was based on metrics readily obtainable with IMUs.
• Linear discriminant analysis (LDA) was employed for classification.

Conclusions:

• Objective, data-driven movement screening is feasible using IMUs and machine learning.
• This method offers a significant advancement for field-based kinematic data collection.
• The developed tool provides an inexpensive and objective solution for enhancing sports screening, assessment, and rehabilitation.