Machine learning-based classification of Taekwondo Poomsae side kick performance using kinematic parameters and...

Area of Science:

• Sports Science
• Biomechanics
• Machine Learning in Sports

Background:

• Taekwondo Poomsae performance assessment traditionally relies on subjective evaluations.
• Objective and quantifiable metrics are needed to improve the accuracy and consistency of performance classification.

Purpose of the Study:

• To develop and validate machine learning (ML) models for classifying Taekwondo Poomsae side kick (SK) performance.
• To identify key kinematic and physical function parameters predictive of SK performance quality.

Main Methods:

• Developed two ML models: one kinematic (SK and pelvic tilt angles) and one physical function (range of motion, Y-balance test).
• Utilized random forest classifiers and evaluated performance using area under the curve (AUC) analysis.
• Forty collegiate Taekwondo athletes performed SKs, with performance assessed by an expert evaluator.

Main Results:

• Both models achieved excellent performance (AUC = 0.930, accuracy = 89.3%).
• SK angle at maximal height was the strongest predictor in the kinematic model.
• Y-balance test composite score was the most impactful parameter in the physical function model.

Conclusions:

• ML models provide accurate and objective classification of Taekwondo SK performance, surpassing traditional subjective methods.
• Kinematic parameters (SK angle) and physical function (dynamic balance) are crucial for performance assessment.
• These findings offer quantitative criteria for evaluating Taekwondo Poomsae side kick execution.