(50) COMPARATIVE VALIDATION AND RELIABILITY OF ARTIFICIAL INTELLIGENCE IN COUNTERMOVEMENT JUMP ASSESSMENT

Purpose: This study aimed to evaluate the intra- and inter-session reliability and validity of an AI-enhanced version of the My Jump 3 application (My Jump 3 AI) compared with established methods—including a criterion force platform, an infrared sensor mat, and manual slow motion video analysis—in male athletes from four sport groups. Methods: Eighty athletes (20 each from Recreational, Basketball, Soccer, and Volleyball groups) performed six maximal-effort CMJs (three per session across two sessions separated by 7–10 days). Jump height was simultaneously recorded using a ForceDecks force platform (1,000 Hz), two slow motion video devices (240 fps) for observer-based My Jump 3, an AI-enhanced My Jump 3 application (30 fps), and an IR Contact Mat. Intra- and inter-session reliability were quantified using intraclass correlation coefficients (ICCs) and coefficients of variation (CVs), while validity was assessed via ICCs and Bland–Altman analyses against the force platform. Results: The force platform, IR Contact Mat, and observer-based My Jump 3 methods demonstrated excellent within-session reliability (ICCs ≥ 0.98; CVs ≈ 3.2–3.5%). The AI-enhanced application showed acceptable reliability (ICCs ranging from 0.86 to 0.96) but with higher measurement variability (CVs up to ~7.5%). Validity analysis revealed that observer-based slow motion video methods closely matched force platform measurements (ICC up to 0.996), whereas the AI-based approach exhibited a modest bias (−3.19 cm) and wider limits of agreement. Notably, volleyball players achieved the highest jump heights across all devices (p < 0.05). Conclusions: While traditional high-speed video analysis and force platforms remain the gold standard for CMJ assessment, the AI-enhanced My Jump 3 application represents a promising alternative. However, improvements in sampling rate and detection algorithms are needed to reduce measurement variability and fully match the precision of established methods. Practical Applications: AI-based CMJ assessment tools offer rapid, portable, and cost-effective solutions for field-based performance monitoring. Practitioners can utilize these technologies for large-scale or remote testing, provided that appropriate calibration and environmental controls are implemented to ensure measurement accuracy. Acknowledgements: We would like to thank the students from Universidad Autónoma de Nuevo León for their invaluable support and dedication throughout this study. Their enthusiasm, perseverance, and commitment greatly contributed to the research. We truly appreciate their hard work and collaboration in making this project a success.