Related Experiment Video
Updated: May 9, 2026
06:52
An Inertial Measurement Unit Based Method to Estimate Hip and Knee Joint Kinematics in Team Sport Athletes on the Field
Published on: May 26, 2020
AI-Powered Monitoring of the Acute: Chronic Workload Ratio: Interpretable Injury Risk Prediction in Soccer Players
Deyu Meng1,2, Meiqi Wei1,2, Shichun He2,3
1School of Athletic Ability Development, Changchun University, Changchun, Jilin, China.
Sports Health
|May 8, 2026
Summary
This study introduces a predictive model for monitoring soccer players' acute:chronic workload ratio (ACWR) using historical data. The model effectively forecasts workload dynamics and identifies elevated injury risks, aiding in optimal training management.
Area of Science:
- Sports Science
- Data Science
- Machine Learning
Background:
- Monitoring the acute:chronic workload ratio (ACWR) is crucial for athlete performance and injury prevention.
- Existing methods may not fully capture the complex dynamics of training load.
- Predicting future ACWR is essential for proactive athlete management.
Purpose of the Study:
- To propose and validate a novel time-series model for predicting soccer players' future ACWR.
- To leverage advanced machine learning techniques, including Transformer-based models and large language models, for enhanced predictive accuracy.
- To provide a practical tool for monitoring athlete workload and identifying potential injury risks.
Main Methods:
- A Transformer-based foundation model, the Tabular Probabilistic Forecasting Network for Time Series, was employed.
- Historical training data, including sensor data (GPS, accelerometers) and subjective feedback, were utilized.
- Prompt engineering and large language models (DeepSeek) were used to extract knowledge-based features and improve predictions.
Main Results:
- The model demonstrated strong performance in ACWR prediction with a low mean absolute error (0.119) and a good R-squared value (0.564).
- For ACWR risk prediction, the model achieved high accuracy (87.12%), precision (85.91%), recall (87.12%), and F1 score (85.27%).
Conclusions:
- The developed model effectively predicts future injury risk in soccer players by analyzing ACWR.
- It serves as a practical tool for managing workload fluctuations and maintaining optimal training states.
- Early identification of elevated injury risk due to abnormal ACWR is facilitated, supporting athlete health and performance.
