Development and Validation of a Kinetics Prediction Model for Football Cutting Using a Single Trunk-Mounted IMU

This study aimed to estimate vertical ground reaction force (vGRF) and lower-limb joint moments during football cutting movements using a trunk-mounted inertial measurement unit (IMU) combined with a Random Forest model, and to validate the feasibility of this approach. IMU data collected during 45° cutting tasks were corrected using an Extended Kalman Filter (EKF). The model demonstrated good and consistent performance for vGRF (coefficient of determination, R² = 0.766; correlation coefficient, r = 0.796) and sagittal plane moments of the ankle and knee (R² = 0.661-0.689, r = 0.807-0.842). While Bland-Altman analysis indicated low bias and generally good agreement, precision at the individual-trial level and accuracy for non-sagittal plane moments somewhat reflected the inherent within-player trial-to-trial variability in movement execution, particularly in non-sagittal loading patterns. It should be noted that performance estimates under the current trial-based validation design may differ from those obtained using a subject-independent framework such as leave-one-subject-out cross-validation. This study demonstrates that a single trunk-mounted IMU can reliably estimate key lower-limb loading patterns, providing a practical foundation for wearable-based kinetic monitoring in applied football settings.