Enhancing Heart Disease Prediction with A Scalable Hybrid Model using Machine Learning

Most difficult to predict in medical data analysis is precisely a risk of heart disease for a patient which lowers death rate or improving survival rates at early stage forecast treatment timely and better resulting than to reach at peak rates . Regularly observing by medical personnel by early detecting cardiac problem can increases the survival of a patient to live more . As doctors can’t continuously able to be in contact with patients , for this we provide decision making based on data collected by various physicians all over the world , Machine learning (ML) help health professionals for predicting and detecting of HD. This targets to provide an accurate algorithm of ML for detecting heart disease at early rate . paper provides an composite approach for accuracy improving . Model that is underlined to remove outliers uses IQR method. Close to 10 models of ML are used. Tomek links with SMOTE also apply with cross validation model to give enhance result as well as ROC values. Eventually we trained as well as proposed a hybrid model that tip over in ratio of 75:25 with correctness of 93% The highest ROC AUC achieved 0.933.