A new open-access study published on 23 October 2026 in BMC Sports Science, Medicine and Rehabilitation reports the development of an interpretable machine learning model designed to predict peak ACL force during side-cutting maneuvers, using biomechanical measurements that can be captured in a laboratory setting. The work, which the authors describe as an assessment method for sports training and monitoring, combines motion analysis, force measurement and muscle activation data within a single predictive framework.
The research team recruited 204 male football and basketball athletes for the study. During side-cutting maneuvers, the researchers extracted kinematic, kinetic and electromyographic parameters using a Qualisys optical motion capture system, force plates and surface electromyography. These measurement streams recorded how the athletes’ joints moved, what forces acted on their bodies and how their muscles responded, providing a multidimensional picture of the biomechanics underlying each cutting action. Machine learning algorithms were then used for model training and testing to predict peak ACL force, and Shapley Additive Explanations, known as SHAP, were employed to interpret the contribution of individual feature values within the model.
The best-performing approach proved to be an Extreme Gradient Boosting, or XGBoost, model. According to the study, this model demonstrated high fitting performance with an R² of 0.929, alongside a low overall deviation, measured as an RMSE of 0.051, and a prediction error, measured as an MAE of 0.032. These figures indicate that the model’s predictions tracked the measured peak ACL force values closely across the test data, with small residual differences between predicted and observed values. The authors report that the machine learning model integrating multidimensional indicators achieved high accuracy and improved the predictive capability of peak ACL force in practical applications.
A central contribution of the study lies in its use of SHAP analysis to open up the model’s decision-making. Global SHAP importance analysis identified knee flexion angle, the hamstring-to-quadriceps activation ratio, hip flexion angle, knee valgus moment, ankle valgus angle, hip adduction angle and anterior tibial shear force as the key features for predicting peak ACL force. The SHAP importance values reported for individual features were 0.07 for knee flexion angle, 0.05 for the hamstring-to-quadriceps ratio, 0.03 for hip flexion angle, and 0.02 each for knee valgus moment and ankle valgus angle, values the authors describe as showing relatively high contributions and strong predictive ability. This ranking suggests that both the posture of the lower limb and the balance of muscle activation between the hamstring and quadriceps groups carry substantial weight in the model’s estimates.
The study’s conclusion frames the underlying mechanism in those terms: the increase in peak ACL force during side-cutting maneuvers is, in the authors’ analysis, essentially the result of the coupling of dynamic postural control mechanical loads and neuromuscular effects. By integrating indicators spanning joint angles, moments, forces and muscle activation ratios, the model provides an interpretation for the comprehensive analysis of peak ACL force rather than relying on any single measurement.
For sports training and monitoring contexts, the findings suggest that data gathered from motion capture, force plates and surface electromyography can be combined into an assessment method that estimates peak ACL force during side-cutting maneuvers. The study’s stated objective was to develop such an interpretable model based on biomechanical measurement parameters, and the reported results present the XGBoost model with SHAP-based feature interpretation as a candidate approach. As with any laboratory-based predictive study, the model’s performance figures describe its behavior on the data used for training and testing, and the authors position the work as providing an assessment method and an interpretation framework for the comprehensive analysis of peak ACL force during side-cutting maneuvers.
Subject of Research: An interpretable machine learning model for predicting peak ACL force during side-cutting maneuvers using biomechanical and neuromuscular measurements
Article Title: Identification and validation of an interpretable machine learning model for predicting peak ACL force during side-cutting maneuvers
Article References: Zhang, Y., Chen, T., Zhu, Y., Yin, Y., Chen, S., Bi, G., & Yu, J. (2026). Identification and validation of an interpretable machine learning model for predicting peak ACL force during side-cutting maneuvers. BMC Sports Science, Medicine and Rehabilitation. https://doi.org/10.1186/s13102-026-02140-7
Image Credits: AI Generated
DOI: 10.1186/s13102-026-02140-7
Keywords: ACL force prediction, machine learning, side-cutting maneuvers, biomechanics, XGBoost, SHAP interpretability, motion capture, force plates, surface electromyography, sports injury monitoring, knee flexion angle, neuromuscular control
News Source: Teresa Odom. (October 29, 2026). Machine learning model predicts peak ACL force during side-cutting maneuvers. Scienmag.
