Erkan Boğa
The European Research Journal - 2025;11(6):1204-1212
Objectives: The aim of this study is to evaluate the effectiveness of the Gradient Boosting algorithm in pre - dicting mortality risk among emergency department patients and to identify the most critical demographic, clinical, and physiological data for these predictions. This study is designed to support early identification and enhance clinical decision support systems. Methods: This retrospective study analyzed data from 1,500 patients who visited a state hospital's emergency department between January 1 and August 31, 2024. Data were collected based on multidimensional features such as demographic information, vital signs, laboratory results, and clinical history. The Gradient Boosting algorithm was used to develop the model, and its performance was evaluated using metrics such as accuracy, sensitivity, specificity, and F1 score. Results: The Gradient Boosting model identified oxygen saturation, age, and heart rate as the most significant predictors of mortality. The CatBoost algorithm demonstrated the highest performance with an accuracy of 88.8% and an F1 score of 85%. The model was proven to be highly accurate in predicting mortality risk. Conclusions: Gradient Boosting algorithms, particularly CatBoost, emerged as a reliable and effective tool for predicting mortality risk. This model can contribute to the development of clinical decision support systems in emergency department settings.
