UCLA

This study examines how machine learning can improve those predictions using real clinical data from 299 patients. Three models were tested: logistic regression, decision tree, and XGBoost. The analysis focused on key health indicators, including age, ejection fraction, kidney function, and sodium levels. After preparing the data and training the models, their performance was measured using several metrics, including accuracy and the area under the ROC curve. Logistic regression achieved the highest accuracy of 85 percent. XGBoost performed best overall in capturing complex patterns, with an AUC of 0.8920. While the decision tree offered an easy to understand approach with good results, achieving an accuracy of 80 percent. Each model had strengths, depending on whether simplicity or predictive power was more important.