Automated detection of atrial fibrillation episode using novel heart rate variability features

Area of Science:

• Cardiology
• Biomedical Engineering
• Signal Processing

Background:

• Atrial fibrillation (AF) is a prevalent arrhythmia impacting millions, often diagnosed through manual ECG analysis, delaying treatment.
• Wearable devices offer continuous RR interval monitoring, enabling automated AF detection.
• Current AF detection methods can be time-consuming and require expert interpretation.

Purpose of the Study:

• To develop novel heart rate variability (HRV) features from RR intervals for AF detection.
• To engineer automated classifiers for timely AF episode identification.
• To compare the performance of proposed classifiers against existing methods.

Main Methods:

• Engineered novel linear and non-linear HRV features from RR intervals.
• Developed and compared automated classifiers for AF detection.
• Utilized data from wearable sensors for feature extraction and classification.

Main Results:

• The proposed automated classifier achieved high sensitivity (98%) and specificity (95%).
• Novel HRV features derived from RR intervals proved effective for AF detection.
• The developed method demonstrated superior performance compared to prior published works.

Conclusions:

• Automated AF detection is feasible and accurate using HRV features from wearable devices.
• The proposed method offers a promising approach for early AF diagnosis and intervention.
• This technology can significantly improve patient outcomes by enabling timely medical care.