On the Beat Detection Performance in Long-Term ECG Monitoring Scenarios

Area of Science:

• Biomedical Engineering
• Cardiology
• Signal Processing

Background:

• Existing R-wave detection algorithms face challenges with long-term Holter monitoring requirements.
• New applications necessitate robust and efficient beat detection methods for extended recordings.

Purpose of the Study:

• To evaluate the limitations of current Holter monitoring algorithms for long-term use.
• To propose and validate a novel beat detection method suitable for long-term scenarios.

Main Methods:

• Longitudinal and transversal studies analyzed widely used waveform analysis algorithms.
• Optimized simultaneous-multilead processing was applied to 7-day Holter monitoring databases.
• A Polling function-based beat detection method was developed and benchmarked.

Main Results:

• The Polling function method achieved high performance metrics: 99.48% sensitivity, 99.54% specificity, 99.69% positive predictive value, and 99.46% accuracy.
• The algorithm demonstrated superior accuracy and computational efficiency compared to other methods.
• Error rate was as low as 0.85% on the MIT-BIH arrhythmia database.

Conclusions:

• The proposed Polling function-based R-wave detection method is accurate and computationally efficient.
• This method is suitable for implementation in long-term Holter monitoring systems.
• It addresses the critical need for reliable noise-avoidance and efficiency in extended cardiac monitoring.