A novel ECG signal classification method using DEA-ELM.

Area of Science:

• Biomedical Engineering
• Computational Cardiology
• Machine Learning in Healthcare

Background:

• Electrocardiogram (ECG) signals are crucial for assessing heart's electrical activity.
• Computer-aided systems enhance ECG analysis for diagnosis and classification.
• Optimizing machine learning models is key to improving diagnostic accuracy.

Purpose of the Study:

• To optimize the number of hidden neurons in Extreme Learning Machine (ELM) for ECG signal classification.
• To enhance the accuracy of ECG signal classification using Differential Evolution Algorithm (DEA).
• To evaluate the performance of the optimized ELM model against conventional approaches.

Main Methods:

• Utilized publicly available ECG records from Physionet.
• Applied Pan-Tompkins Technique (PTT) and Discrete Wavelet Transform (DWT) for feature extraction (PR, QT, ST periods, QRS wave).
• Implemented Extreme Learning Machine (ELM) for classification and Differential Evolution Algorithm (DEA) to optimize ELM's hidden neuron count.

Main Results:

• The optimized ELM model achieved a highest classification accuracy of 97.5%.
• The optimal number of hidden neurons determined by DEA was 93.
• The DEA-improved ELM demonstrated superior performance compared to the conventional ELM.

Conclusions:

• Optimizing hidden neuron count in ELM using DEA significantly improves ECG signal classification accuracy.
• The proposed method offers a more accurate and efficient approach to ECG analysis.
• This technique holds potential for advancing computer-aided diagnosis in cardiology.