Design of a Prototype Application for Predicting Death Due to Heart Failure Using Machine Learning Methods Based on Heart Failure Clinical Records Data

Abstract

This research aims to develop a prototype of the Heart Failure Death Prediction Application using machine learning methods based on clinical data from the Heart Failure Clinical Records. The application utilizes clinical patient data, such as age, blood pressure, ejection fraction, creatinine levels, and other attributes, to build a predictive model for mortality risk. Several machines learning algorithms, including Random Forest, Logistic Regression, and K-Nearest Neighbors (KNN), were employed to model and analyze the data. The dataset used in this study consists of 299 clinical records with 13 attribute columns. The target attribute is Death Event, while other attributes, such as age, gender, medical history (anemia, diabetes, high blood pressure), and laboratory test results (creatinine, sodium, and ejection fraction), were used as predictors. The application is equipped with several main menus to support its functionality, such as the Dashboard, which provides a summary of statistical prediction information and related reports, and Blog/News, which offers heart health education. The Data Master menu allows for the management of supporting data, while the Diagnosis menu is used to perform predictions based on patient input data. The Diagnosis History menu stores previous prediction results, while the Patient Data menu facilitates the management of patient information.