Detection of Long Term Variations of Heart Rate Variability in Normal Sinus Rhythm and Atrial Fibrillation ECG Data

Abstract

Circadian variations of heart rate variability (HRV) have been well known in cardiac diseases. However, long term HRV features were not thoroughly investigated for the prediction of atrial fibrillation (AF). Thus, we analyzed the 15 hour long changes of HRV of normal sinus rhythm (NSR) and AF data. Long term patterns of HR V in NSR were established first and normal data of AF were shown different to NSR. Due to long term changes of HRV patterns, two formulas were provided higher accuracy (93%) than single formula (79%) in detecting normal data of AF. Furthermore, HRV features representing 5 and 30 min before the AF onset showed significant temporal changes and the dynamics of these changes were also different depending on the recording periods. These data suggest that the onset of AF could be predicted more accurately by considering the long term temporal variations of HRV features.