Firdausi Kahfi Maulana, Ghulam Zaky Ismail, Susi Susanti, Tita Rachma Ayuningtyas, Isma Nur Azzizah, Azizati Rochmania
Injuries among athletes pose a significant challenge in the world of sports, affecting not only physical performance but also long-term health and team dynamics. The ability to predict injury risk early is crucial in implementing preventive strategies. This study aims to develop predic-tive models for athletic injuries using machine learning algorithms, with a focus on comparing the performance of Logistic Regression and Random Forest on two datasets with differing class distributions, one balanced and one imbalanced. The first dataset (Injury Prediction Dataset) con-tains demographic and physiological data and has a balanced distribution between injured and non-injured classes. The second dataset (Collegiate Athlete Injury Dataset) contains performance and fatigue-related metrics, with a naturally imbalanced distribution where injury cases are rare. Both datasets underwent preprocessing, including feature scaling and label encoding, followed by model training and evaluation using accuracy, precision, recall, and F1-score metrics. Results show that Logistic Regression consistently outperforms Random Forest in detecting minority in-jury cases, especially in the imbalanced dataset, achieving 100% recall. In contrast, Random Forest fails to identify any injury cases despite high overall accuracy. This highlights the importance of model selection and handling class imbalance when building predictive systems for medical or safety-critical domains. The findings demonstrate that simple, well-tuned models like Logistic Regression can be effective for early injury risk prediction, especially when supported by appropriate class balancing techniques. This research contributes to the development of AI-based medical diagnostic tools in the sports domain, offering a foundation for further enhancement us-ing richer datasets and advanced sampling methods. © 2026 by the authors.
Department of Fisioterapi, Universitas Negeri Surabaya, Surabaya, Indonesia; Department of physical education and health recreation, Universitas Negeri Surabaya, Surabaya, Indonesia
