Artificial Intelligence (AI)-Predicted Joint Stress Indices and Overuse Injuries in Martial Arts Training Among Martial Arts Athletes

Abstract

Artificial intelligence  offers a paradigm shift from reactive to proactive injury prevention in sports, yet its practical acceptance and accuracy in high-impact disciplines like martial arts require empirical validation. This study investigated martial arts athletes' perceptions of AI-predicted joint stress indices and their correlation with overuse injuries, guided by the Human-AI Teaming framework. Employing a descriptive-comparative-correlational design, 355 martial arts athletes from a university setting completed a validated questionnaire assessing AI-predicted indices across five dimensions: stability, symmetry, alignment, variability, and reproducibility. Descriptive and inferential statistics analyzed the data.Results indicated a uniform, moderate perception of the AI system's accuracy, with an overall mean rating of 1.91 (SD=0.79) on a 4-point Likert scale, interpreted as "Slightly Accurate." Reproducibility of feedback was the highest-rated dimension (Mean=1.94), while stability, symmetry, and alignment were rated lowest (Mean=1.90). Statistical analyses revealed no significant differences in these perceptions based on athlete sex, age, martial arts discipline, or years of experience. The findings suggest that while AI is recognized as a potentially valuable tool for longitudinal monitoring, its current application is perceived as lacking the nuanced accuracy required for detailed biomechanical feedback and immediate technical correction. The study concludes that AI-predicted joint stress indices hold promise as a supplementary tool for injury prevention but require significant algorithmic refinement to improve biomechanical fidelity and athlete trust before achieving full integration into personalized training frameworks.