Performance Evaluation of Machine Learning Algorithms for Heart Disease Prediction Using Real-World Data from Yemen

Heart disease remains one of the leading causes of death globally. To better detect heart illness, a number of researchers have suggested developing accurate prediction algorithms. Medical costs are reduced, lives are saved, and the likelihood of human error is reduced using these procedures. The dataset used in this study was collected in 2023 from Aden, Yemen’s Algomhoryah Hospital and pertains to cardiac diseases. This study employed a data mining strategy from machine learning, an area of artificial intelligence, to forecast and identify cardiac conditions. The data was cleaned and discretized the data after collection, among other preparation procedures. A training set and a test set were created from this dataset using the Weka cross-validation tool. To assess how well each model worked, the study used measures such as accuracy, sensitivity, specificity, and mean absolute error (MAE). For the purpose of cardiac disease classification, four algorithms were employed: K-nearest neighbour (KNN), Naive Bayes (NB), Support vector machine (SVM), and Decision Tree (DT) J48. In the studies, the Decision Tree (J48) classifier achieved an accuracy of 91.30%, while the K-nearest neighbour (KNN) classifier achieved 91.00%.