Injury prediction in athletes remains a significant challenge in sports medicine, where early detection is crucial for preventing severe injuries and optimizing performance. This research investigates the application of machine learning (ML) techniques to predict injury risk based on key physiological parameters, including heart rate, impact force, activity duration, and skin temperature. A dataset comprising recorded athlete data under various conditions is analyzed to determine the predictive significance of these features. While traditional classifiers exhibited limitations due to the non-collinear nature of the feature set, boosting techniques emerged as the most effective approach, achieving the highest predictive accuracy. The study evaluates multiple ML methodologies, including boosting algorithms, neural networks, and conventional classifiers, highlighting their comparative performance. Among these, LightGBM demonstrated the best performance when applied to polynomial degree 2 transformations with Recursive Feature Elimination (RFE)-selected 9 features, achieving an accuracy of 76.32%. By identifying intricate patterns and correlations within the data, this research facilitates accurate, real-time injury predictions, enabling coaches and healthcare professionals to proactively intervene and mitigate injury risks. This study contributes to the field of Integrating Machine Learning into Software Systems by developing an AI-powered predictive model that can be seamlessly incorporated into sports tracking and health monitoring applications. The proposed framework also helps showcase how advanced machine learning techniques can enhance software-driven decision-making processes in sports science. By leveraging ML-powered injury prediction models, sports organizations can enhance athlete longevity, reduce rehabilitation costs, and optimize training regimens, ultimately advancing sports medicine through personalized and precise injury prevention strategies.