Sports Injury Risk Prediction and Intervention for College Students Based on Decision Tree Algorithm

Abstract

Sports injuries pose significant risks to elite athletes' careers and performance, necessitating accurate predictive models for proactive prevention strategies. This study develops an interpretable machine learning approach for early injury risk assessment using data from 108 elite athletes across multiple sports disciplines, including training load metrics, injury history, physiological parameters, and lifestyle factors. Four machine learning algorithms - Decision Tree, Random Forest, Support Vector Machine (SVM), and Logistic Regression - were trained and evaluated using 5-fold cross-validation. The Decision Tree model achieved superior performance with 82.4% accuracy, 85.3% recall, 78.6% precision, and an AUC of 0.876, outperforming Random Forest (80.2%), SVM (76.8%), and Logistic Regression (74.1%). Feature importance analysis revealed that injury count in the past 6 months (0.243), consecutive training days (0.186), and training intensity (0.152) collectively account for 58.1% of the model's predictive power. Extracted decision rules provide actionable guidance: athletes with ≥2 recent injuries training ≥6 consecutive days face 91.2% injury probability, while those with <2 injuries and training intensity ≤85% show only 23.7% risk. The dominance of modifiable training-related factors over demographic variables suggests that targeted workload management and recovery optimization can significantly reduce injury incidence. This interpretable model transforms injury prevention from reactive treatment to proactive risk management, providing coaches and medical staff with transparent, evidence-based decision support for elite sports environments.