This study investigates the application of supervised machine learning methods for predicting injury outcomes among participants involved in road traffic incidents. The analysis is based on detailed participant-level data collected in Poland between 2015 and 2022, covering over six million records. The dataset includes individual and incident-related characteristics, such as participant role, gender, driving license status, legal responsibility, and area type. A multi-class classification framework was developed to predict the injury status of participants, categorized as no injury, light injury, severe injury, or fatality. Three machine learning models - Random Forest, XGBoost, and LightGBM - were implemented and evaluated in terms of predictive performance. In addition, an analysis of feature importance was conducted to identify the most influential factors contributing to injury severity. The results demonstrated that ensemble learning models, particularly XGBoost, achieved the highest predictive performance. Participant role and legal responsibility were identified as the most critical factors influencing injury outcomes. The findings confirm the potential of machine learning techniques to improve the understanding of individual-level determinants of injury severity and to support data-driven road safety policies.