MI-CSBO: a hybrid system for myocardial infarction classification using deep learning and Bayesian optimization.

Area of Science:

• Cardiology
• Biomedical Engineering
• Machine Learning

Background:

• Myocardial Infarction (MI) is heart tissue damage from blocked coronary arteries, often due to atherosclerosis.
• Risk factors include hypertension, diabetes, high cholesterol, and genetic predisposition.
• Early and accurate MI detection and classification are critical for patient outcomes.

Purpose of the Study:

• To introduce a new hybrid approach, MI-CSBO, for classifying Myocardial Infarction using Electrocardiogram (ECG) data.
• To enhance the diagnostic accuracy of MI detection through advanced signal processing and machine learning.

Main Methods:

• ECG signals from the PTB Database were transformed into spectrograms (frequency domain).
• A deep residual Convolutional Neural Network (CNN) was applied to the ECG spectrograms.
• Bayesian optimization, NCA feature selection, and various machine learning algorithms (k-NN, SVM, Tree, Bagged, Naïve Bayes, Ensemble) were employed for classification.

Main Results:

• The MI-CSBO method demonstrated a 100% correct diagnosis rate for Myocardial Infarction.
• The hybrid approach effectively integrated time-frequency analysis with deep learning and Bayesian optimization.

Conclusions:

• The MI-CSBO approach offers a highly accurate and reliable method for MI classification from ECG data.
• This technique holds significant potential for improving the early diagnosis and management of heart attacks.