A Machine Learning Based Predictive Model to Diagnose Heart Failure Patients using Imbalanced Classification Problem

Heart failure (HF) is now one of the most common diseases, causing approximately seventeen million death cases every year all over the world. HF occurs due to less pumping ratio of blood by the heart that a normal human being needs to survive. In this regard, research studies have been proposed to predict the causes behind the heart failure of a patient using the 'Heart failure clinical records (HFCR's) dataset. Although, much research has been performed on this dataset, however, there is a lack of construction of a more reliable predictive model that helps to predict HF patients with better prediction results. We aimed to apply imbalance learning to handle the imbalance dataset as a very few researchers applied it. We trained the models using six ensemble and three non-ensemble classifiers with the help of multiple experiments. In the end, we performed an evaluation measure to compare our prediction results with the previous research work. Our proposed model gives a significant increase in accuracy value as well as in precision, recall, and f1-score.