Fatigue-Related and Timescale-Dependent Changes in Individual Movement Patterns Identified Using Support Vector...

Area of Science:

• Human movement science
• Sports biomechanics
• Machine learning applications in sports

Background:

• High-performance sports demand individualized analysis in human movement science.
• Machine learning effectively identifies unique athlete movement patterns and intra-individual changes.
• Understanding movement adaptations during fatigue is crucial for training optimization.

Purpose of the Study:

• To analyze biomechanically described movement patterns during fatigue accumulation in expert karate athletes performing repetitive ballistic kicks.
• To determine if individual movement characteristics persist despite fatigue.
• To investigate how intra-individual movement patterns adapt with increasing fatigue.

Main Methods:

• Sixteen expert karatekas performed 606 frontal foot kicks (mae-geri) at 80% intensity, with maximal intensity kicks interspersed.
• Three-dimensional full-body kinematic data were captured using 10 infrared cameras.
• Supervised machine learning (support vector machine) classified normalized joint angle waveforms.

Main Results:

• Machine learning successfully distinguished unique kinematic movement patterns for each individual athlete.
• Individual movement pattern identification remained consistent regardless of intensity or fatigue levels.
• Adaptations in intra-individual movement patterns were detected throughout the training session, suggesting pattern-based changes rather than increased variance.

Conclusions:

• Individual movement patterns in expert athletes are identifiable even under fatigue conditions.
• Fatigue-related processes induce detectable adaptations in intra-individual movement patterns.
• These findings have implications for personalized training strategies and performance analysis in sports.