Heart Disease Prediction and Classification Using Machine Learning Algorithms

Machine learning has been employed to develop a high-performance prediction model for heart disease. Indeed, various data can be used to predict heart disease using machine learning, but some of them are imbalanced data. Therefore, these imbalanced data require special treatment. In this work, we used mutual information as a feature selection to find the best features in our data, we employed the SMOTE(Synthetic Minority Oversampling Technique) to balance the data because our dataset was imbalanced, which can be biased our models’ performance, and we utilized some machine-learning techniques such as k-nearest-neighbors, support vector machine, random forest, decision tree, Naïve Bayes, artificial neural network, and logistic regression. This dataset is the UCI dataset, which was downloaded from Kaggle and has been used during this research. The experimental results show that k-nearest neighbors achieve a better prediction performance than the others, which provided an accuracy of 0.91, a precision of 0.91, a recall of 0.94, a F1-score of 0.92, and an AUC-ROC value of 0.906.