This study evaluated machine learning (ML) models for predicting mortality and respiratory failure in trauma patients with rib fractures. The dataset included 260,771 patients from the National Trauma Data Bank. Models were based on light gradient boosting machine (LGBM) and extreme gradient boosting algorithms, and k-fold cross-validation was used to evaluate the models. The model was designed to mimic real-life clinical scenarios by using only admission variables and testing robustness to missing data. LGBM achieved accuracies of 0.97 for mortality and 0.93 for respiratory failure, outperforming other tested models. Age, Glasgow coma score, and heart rate were the most influential predictors of mortality. The limited performance of previous clinical tools underscores the need for more accurate measures, such as ML models. These models can detect complex patterns that traditional tools may overlook.