Between-system agreement of K-AI wearable tech (K-sport) and vector S7 global navigation satellite systems for measuring distance, velocity, and acceleration metrics

This study evaluated the agreement between K-AI Wearable Tech and the previously validated Vector S7 during an Australian football training session. Thirty sub-elite Australian football male players (21.6 ± 2.8 years) took part in the study. The K-AI Wearable Tech demonstrated mean absolute error (MAE) and root mean square error (RMSE) values of 74.1 m (1.00%) and 133.9 m (1.68%) for total distance, 52.6 m (6.46%) and 64.7 m (10.42%) for combined high-speed running-sprinting distance (> 19  km ⋅ h − 1 ), 6.5 m (4.66%) and 9.8 m (6.51%) for sprinting distance (> 25  km ⋅ h − 1 ). Acceleration and deceleration counts were 5.9 (13.92%) and 7.6 (17.04%), and 2.3 (10.07%) and 3.0 (13.66%), respectively. Peak and instantaneous velocity showed MAE and RMSE of 0.35  km ⋅ h − 1 (1.20%) and 0.39  km ⋅ h − 1 (1.35%), and 0.11  km ⋅ h − 1 (2.26%) and 0.15  km ⋅ h − 1 (3.03%), respectively. Mean biases (95% limits of agreement) were 64.2 m (−170.0 to 298.4) for total distance, 48.6 m (−36.6 to 133.8) for combined high-speed running-sprinting distance (> 19  km ⋅ h − 1 ), and −4.0 m (−21.9 to 13.9) for sprinting distance (> 25  km ⋅ h − 1 ). Acceleration and deceleration counts showed biases of −3.1 (−17.0 to 10.7) and 1.0 (−4.6 to 6.7), while peak and instantaneous velocity were 0.34  km ⋅ h − 1 (−0.05 to 0.73) and 0.11  km ⋅ h − 1 (−0.09 to 0.32), respectively. The intraclass correlation coefficient (ICC) ranged from very large to nearly perfect across all metrics assessed. Collectively, these findings demonstrate that the K-AI Wearable Tech produced distance-, velocity-, and acceleration-based outputs comparable to those obtained from Vector S7. However, the magnitudes of both absolute and proportional differences should be considered when interpreting agreement between systems.