A Personalized Arrhythmia Monitoring Platform.

Area of Science:

• Biomedical Engineering
• Cardiology
• Signal Processing

Background:

• Arrhythmia detection is crucial for cardiovascular disease diagnosis but lacks generic real-time solutions due to ECG signal variability.
• Automated classification of arrhythmias relies heavily on effective feature extraction and classification techniques.

Purpose of the Study:

• To develop a personalized arrhythmia monitoring platform for real-time detection of arrhythmias from ECG signals.
• To enable point-of-care cardiovascular disease diagnosis through advanced signal analysis.

Main Methods:

• Employed the discrete orthogonal stockwell transform (DOST) for time-frequency feature extraction from ECG signals.
• Utilized an artificial bee colony (ABC) optimized twin least-square support vector machine (LSTSVM) for feature classification.
• Optimized feature set dimensionality and classifier parameters using ABC.

Main Results:

• The proposed method achieved high accuracy in classifying ECG signals: 96.29% for the class scheme and 96.08% for the personalized scheme.
• The system was prototyped on an ARM-based embedded platform and validated on the MIT-BIH arrhythmia database.
• Performance surpassed existing state-of-the-art methods in cardiovascular disease diagnosis.

Conclusions:

• The developed personalized arrhythmia monitoring platform demonstrates significant potential for accurate, real-time ECG analysis.
• The novel DOST and ABC-LSTSVM approach offers an effective solution for automated arrhythmia detection.
• This technology can enhance point-of-care cardiovascular diagnostics and patient management.