Heart Failure Disease Prediction Using Machine Learning Models

Abstract

Heart failure disease affects 26 million people worldwide and it has a lower survival rate than breast or prostate cancer. An early diagnostic of the disease is very important for prevention and possible treatment. In this work, we propose using machine learning models to predict the probability of developing a heart failure disease in a patient. We compare two machine learning models over a public dataset of risk factors and patients’ clinical features. After a comparative analysis, we find that a logistic regression model can predict 87% of the cases on the data base. After that, we implement an easy web application for heart failure disease prediction. We anticipate that applying this model hospitals will be able to reduce their patient admission due to heart failure disease and patients will be able to reduce their risk and avoid all the implicit costs.