Abstract:
Insufficient blood circulation throughout the body is referred to as heart failure. The heart
normally becomes too weak or stiff to function properly when this occurs. Occasionally,
congestive heart failure is used to describe it. Congestive heart failure life expectancy
varies with disease severity, age, genetics, and other variables. As per the Centers for
Disease Control and Prevention (CDC), nearly half of congestive heart failure patients
survive for more than five years. Several studies have attempted to forecast the survival
rate of patients with heart failure. However, most of these studies have utilized a dataset
consisting of 299 rows and have failed to balance the dataset. The primary goal of this
research is to build an explainable machine learning model that can predict the survival
status of heart failure patients using machine learning algorithms. In this research, we used
machine learning algorithms for prediction. Machine learning serves as a means for
Artificial Intelligence systems to accomplish a multitude of tasks, among them being the
prediction of output values from input data. We used the latest machine learning
algorithms. The data is collected from Felege Hiwot referral hospital and Injibara general
hospital. The algorithms used for prediction model are DT, LR, KNN, DNN, and XGBoost.
According to the experimental findings, XGBoost outperforms other machine learning
techniques by producing an AUC value of 0.93 using Grid Search HPO after SMOTE.
Keyword: - Heart Failure; Survival Event; Follow-up Period; Synthetic Minority
Oversampling Technique; Hyper-parameter Optimization; Model Explainability