DETECTION OF MYOCARDIAL INFARCTION CARDIAC DISEASES BASED ON RENYI ENTROPY AND OPTIMUM BLOCK LENGTH OF ONLINE SEQUENTIAL-EXTREME LEARNING MACHINE

Myocardial Infarction (MI), or heart attack, is an extremely severe cardiac condition marked by the partial disruption of blood flow to the myocardium. This can lead to a heart attack and potentially fatal consequences. Early and accurate detection of MI is pivotal in enhancing cardiac function and improving patient outcomes. Detecting MI diseases using Electrocardiogram (ECG) signals presents significant challenges, particularly in accurately extracting patterns like QRS complexes and ST segments. This process demands sophisticated algorithms to analyze complex EEG data while ensuring high detection accuracy and computational efficiency. This study proposes an efficient automated approach for detecting MI, utilizing Stationary Wavelet Transform (SWT), Renyi entropy (RenyEn), and Optimized Block Length Online Sequential-Extreme Learning Machine (OBL-OS-ELM) algorithm, with and without Kernel Principal Component Analysis (KPCA). The SWT is utilized to decompose the ECG signal into sub-bands, allowing for the extraction of RenyEn features. The OBL-OS-ELM classifier is utilized to classify the health status of individuals, differentiating between normal (NR) subjects and ten categories of MI based on features derived from KPCA as well as those obtained without KPCA. The findings of the study demonstrate that the method incorporating KPCA achieving an Accuracy (Ac) of 99.48%, Sensitivity (Se) of 99.60%, Specificity (Sp) of 99.98%, and a Positive Predictive Value (PPV) of 99.99%. These results were attained with an optimal block length of 40 and a hidden layer size of 200, utilizing the sigmoid activation function.