The study was concerned with the development of a prediction model for the mortality of critically ill patients with coronary artery disease (CAD) in the intensive care unit (ICU). Data from two databases (MIMIC-IV and MIMIC-III) and machine learning methods were used. The aim was to predict the 28- and 365-day risk of mortality. The RandomForest model proved to be the best at predicting both short-term and long-term mortality risk. The model was validated both internally and externally and achieved high accuracy (AUC up to 0.914). The study identified important risk factors using SHAP analysis. The results suggest that this model can help doctors identify patients at high risk of death. However, further research is needed to confirm the clinical utility of the model.