Application of Machine Learning Techniques in Sepsis Mortality Prediction

Sepsis, a life-threatening condition resulting from an infection, remains one of the leading causes of mortality in critically ill patients. Early identification of high-risk patients is essential for timely intervention, which can significantly improve survival rates. This study explores the use of machine learning (ML) models, specifically XGBoost, Random Forest, and a Stacked ensemble model, for predicting sepsis-related mortality using the MIMIC-III critical care database. The research compares the performance of these models based on accuracy and AUC-ROC scores. After hyperparameter tuning, XGBoost achieved the highest performance with an AUC of ۸۳.۹۶% and an accuracy of ۸۵.۲۹%. This paper demonstrates the potential of machine learning models as effective tools for supporting clinical decisions in sepsis mortality prediction.

کلیدواژه ها:

Sepsis ، Machine Learning Models ، MIMIC III ، Mortality Rate ، Prediction