Machine learning classification models for the patients who have heart failure

Abstract

Heart failure is a cardiovascular disease with significant morbidity and mortality, affecting a growing number of people worldwide [1]. The aim of this paper is to predict the probability of survival of patients by looking at their various characteristics, diseases, and lifestyles in the most successful way by using various machine learning methods. The 299 patients in the data set we use, had left ventricular systolic dysfunction in 2015 and are classified as New York Heart Association (NYHA) class III and IV. The probability of survival of patients is estimated by applying various machine learning methods on the data set. In this study, there are two versions. In the first version of the study, Principal Component Analysis (PCA) is used to reduce the size of the data set. The performance of the machine learning algorithms is then evaluated using a variety of metrics. In the second version, the data set is only subjected to machine learning techniques, and performance is then assessed. Accuracy, Matthews correlation coefficient (MCC), sensitivity, specifity, F1 score, receiver operating characteristic-area under the curve (ROC-AUC), and precision-recall area under the curve (PR-AUC) values are calculated to measure success. Comparing the two versions reveals that all machine learning algorithms in general have performed better in the second version without PCA. In the second version, the CatBoost algorithm gave the most successful result. Patients with heart failure can have their mortality status predicted using machine learning techniques. The goal of this paper is to look at a variety of characteristics in order to assess the patient’s mortality status. The condition of the patient can be improved by selecting the proper treatment based on the mortality situation.