Coronary Artery Disease prediction using Machine Learning Techniques

Abstract

Cardiovascular disease prediction in the field of clinical record diagnosis is difficult. In the healthcare sector, there is a massive volume of data. Machine learning helps individuals make better choices and forecasts by transforming massive amounts of raw information from the healthcare system into realities. The existing study employed a wide range of machine learning based are used to forecast coronary problems including logistic models, decision trees, neural networks, and so on. The proposed system detects coronary heart disease based on boosting methods, a machine learning methodology. The dataset considered for the work is Framingham datasets with 4238 instances and 14 attributes. To improve the system's efficiency, a Feature- Selector optimization model that included a recursive feature elimination and Boruto method choose the best subset of coronary heart disease traits. Then random over-sampling and SMOTE, an extremely effective optimized model-based technique to handle a problem of imbalanced data. For classifications, Random forest, Decision tree, gradient boosting, Adaptive boosting, and Support vector model are performed. The system seems to have an accuracy of 88% for the recursive feature elimination method with the random forest model.