A novel method for EKG anomaly detection based on the double sliding window technique.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Cardiology

Background:

• Traditional anomaly detection in ECG signals is often computationally intensive.
• Existing methods may require prior data distribution knowledge or fixed thresholds, limiting real-time application.
• Machine learning models can be resource-heavy for embedded or portable systems.

Purpose of the Study:

• To develop an efficient and adaptive anomaly detection technique for ECG signals.
• To enable real-time heart rhythm deviation identification without prior data assumptions.
• To assess the performance of a novel double-sliding-window method.

Main Methods:

• A double-sliding-window approach was implemented for adaptive signal analysis.
• Two independent sliding windows dynamically tracked ECG signal variations.
• The system was evaluated for computational efficiency (execution time, memory usage) and detection accuracy.

Main Results:

• The detection module demonstrated an average execution time of 0.03 seconds and used less than 16 MB of memory.
• Achieved high performance metrics: 95.33% accuracy, 95.00% sensitivity, 100.00% precision, and 100.00% specificity.
• The method proved suitable for real-time systems due to its low resource requirements.

Conclusions:

• The double-sliding-window technique provides an efficient and accurate solution for ECG anomaly detection.
• This adaptive method is well-suited for real-time applications, including portable and resource-constrained devices.
• The study highlights the potential of dynamic signal analysis for improved cardiac monitoring.