An Enhanced Predictive Model for Heart Disease Diagnoses Using Machine Learning Algorithms

Abstract

Heart disease is the primary reason for death in humans, with cardiovascular disorders being responsible for most cases. Risk factors include unhealthy lifestyle, depression, high blood pressure, and high cholesterol. Despite advancements in expert systems for diagnosing heart disease, accurately predicting the disease remains a challenge. To deal with such challenges, the concepts of Machine Learning, and Artificial Intelligence are investigated to provide an overview of the most often deployed methodologies for diagnosing cardiac disease. Cardiovascular diseases can be predicted by investigating the patient's data. In this paper to attain maximum accuracy, ensemble learning is used. Performing pre-processing techniques such as one hot encoding, normalizing the data, removing outliers, and optimizations of models by hyper-parameter, using ensemble learning led our model to an accuracy of 96.7%. In ensemble learning distinctive algorithm are used as weak learners. Every classifier is trained over the dataset and tested to evaluate the performance of each week’s learner. Then the best leaner amongst classifiers is selected for the final model for predictions in the future resulting in introspectively accurate outcomes. This model does not only help the medical community in the diagnosis and curing of heart patients but provides the bases for another researcher to adopt the technique and further improve the model.