[Research on malignant arrhythmia detection algorithm using neural network optimized by genetic algorithm].

Area of Science:

• Biomedical Engineering
• Artificial Intelligence in Medicine
• Cardiology

Context:

• Automated external defibrillators (AEDs) require accurate malignant arrhythmia detection for effective treatment.
• Existing algorithms for arrhythmia classification have limitations in reliability and accuracy.
• Cardiac arrest survival rates are significantly impacted by timely and appropriate defibrillation.

Purpose:

• To develop and evaluate a novel algorithm for the detection and classification of malignant arrhythmias.
• To improve the performance of automated external defibrillators (AEDs) in rhythm analysis.
• To enhance the accuracy of differentiating critical heart rhythms like ventricular fibrillation and ventricular tachycardia.

Summary:

• A backpropagation neural network, optimized by a genetic algorithm, was developed using 21 extracted metrics from existing algorithms.
• The algorithm was trained and tested on 1,343 electrocardiogram (ECG) samples, classifying sinus rhythm, ventricular fibrillation, ventricular tachycardia, and asystole.
• The proposed algorithm achieved a balanced accuracy of 99.06% on the test dataset, outperforming existing methods.

Impact:

• The developed algorithm demonstrates superior performance in malignant arrhythmia detection compared to current approaches.
• Implementing this algorithm in AEDs can increase the reliability of pre-defibrillation rhythm analysis.
• Improved AED functionality is expected to contribute to higher survival rates for patients experiencing cardiac arrest.