Date of Publication :21st May 2024

Abstract:MI is a critical cardiovascular condition that leads to millions of deaths worldwide. Early and accurate diagnosis of MI is crucial to prevent adverse outcomes such as heart failure and fatality. Traditional diagnostic methods, such as angiography, are invasive, costly, and may have associated risks. Therefore, researchers have turned to MLand data mining techniques to develop alternative diagnostic approaches. This paper proposes ELM for the classification of MI severity. To find the most informative subsets of features, the feature ranking algorithms namely MRMR, Relief-F, and Fisher are used on the publicly available MI dataset from the UCI ML Repository. The proposed model is evaluated using a 10-fold cross-validation technique. Comparative analysis is conducted to assess the performance of the ELM classifier with SVM, Random Forest, and XG Boost robust classifiers. The results show the highest accuracy achieved by the ELM model is 99.74% with 40 features selected using the MRMR feature ranking algorithm. However, the highest accuracy achieved by SVM, RF, and XG Boost is 95.09%, 94.1%, and 94.7% respectively. Overall, the proposed ELM model with feature ranking algorithms offers an effective and efficient solution for the severity classification of MI.

Reference :