Abstract:
Heart disease is the top cause of mortality in every country on the planet, according to the World Health Organization. Every year, hundreds of individuals all over the world lose their lives as a result of this disease. Due to the fact that the procedure is both timeconsuming and costly, physicians may still predict heart illness with reasonable accuracy. As a result, the authors propose a technique to help physicians in diagnosing and making better decisions by anticipating the beginning of cardiac disease in order to aid them in their decisions. This research utilizes risk factor data collected from medical records to train five distinct models using five different machine learning techniques. It was decided to utilize Logistic Regression, K-Nearest Neighbours, Gaussian Naive Bayes, Decision Tree, and AdaBoost as the best method for predicting heart disease based on the dataset. A number of variables are considered in order to choose the most appropriate algorithm. With the UCI dataset, the greatest accuracy was achieved by Naïve Bayes with 90.32% accuracy,
followed by KNN and Logistic Regression with 87.1% accuracy. In this article, the accuracy is predicted, and additional variables such as the Jaccard Score and the Cross
Validated Score are shown. The authors conclude by recommending that various validation techniques be used on prospectively acquired data in order to support the suggested methodology.