pRR30, pRR3.25% and Asymmetrical Entropy Descriptors in Atrial Fibrillation Detection.

Area of Science:

• Cardiology
• Biomedical Engineering
• Signal Processing

Background:

• Early detection of atrial fibrillation (AF) is crucial for preventing stroke.
• Heart rate variability (HRV) parameters are used for AF detection.
• pRR30 and pRR3.25% previously showed strong performance in distinguishing AF from sinus rhythm (SR).

Purpose of the Study:

• To compare the diagnostic properties of pRR30, pRR3.25%, and asymmetric entropy for AF detection.
• To evaluate the added value of asymmetric entropy to existing HRV parameters.

Main Methods:

• RR intervals were extracted from 60-second ECG segments from the Physionet Long-Term Atrial Fibrillation Database.
• Analysis included receiver operator curve analysis, confusion matrix, and logistic regression.
• Diagnostic properties of pRR30, pRR3.25%, and asymmetric entropy (H) were assessed.

Main Results:

• The combined model (pRR30, pRR3.25%, H) achieved an AUC of 0.98, slightly higher than pRR30 (0.959) and pRR3.25% (0.972).
• The practical difference in diagnostic performance between individual parameters and the combined model was negligible.
• Combining parameters significantly increased false-negative cases threefold.

Conclusions:

• Asymmetric entropy shows potential for differentiating AF from SR in short ECG segments.
• Adding asymmetric entropy to pRR30 and pRR3.25% does not significantly improve AF detection compared to pRR3.25% alone.
• pRR3.25% remains a highly effective parameter for AF detection.