We developed an IoT and machine learning (ML) system to predict injuries and monitor performance in athletes by integrating advanced sensor technology. IoT devices, including chest straps with heart rate monitors, inertial measurement units, accelerometers, and GPS trackers, were used to collect real-time physiological and biomechanical data. The data collected was analyzed using statistical methods and ML algorithms (Logistic Regression, Random Forest, and Extreme Gradient Boosting). The results showed that training load and fatigue are the most significant predictors of injury risk. While heart rate functioned as an independent marker, the participants under high strain showed significant cardiovascular overexertion with heart rate variability peaking between 100 and 200 BPM and median rates of 140 BPM. Ensemble ML models demonstrated exceptional predictive accuracy, reaching an area under the curve of 1.066. The results of this study demonstrate that the seamless integration of wearable sensors and data-driven analytics offers a robust approach to personalizing training and optimizing injury prevention.