COMPARATIVE ANALYSIS ON SURVIVAL PREDICTION OF HEART FAILURE PATIENTS IN THEIR FOLLOW-UP PERIOD

Abstract

When your heart cannot adequately pump blood to your body's organs, heart failure occurs. If it's untreated, It will result in a person's death. Therefore, a prediction analysis is performed to determine whether a person has a significant chance of surviving heart failure. Machine learning methods such as Logistic Regression, KNN, Decision Tree Classifier, Random Forest Classifier, and Extra Trees Classifier are applied to the dataset which was gathered during the patient's follow-up period. This research paper provides every pre-processing approach and strategy used to produce the most precise model, as well as a Flask implementation and data visualization tools for a deep understanding of the relevant data. We note that an accuracy of 86% is produced by the Extra Trees Classifier and conclude that the Extra Trees Classifier is the most accurate model for survival prediction among all the other models. We utilize Flask to display the outcome of our survival prediction and help to make those predictions understandable to the user. This research can be used to generate accurate predictions of whether or not a heart failure patient will survive by adding more features and improving our model.