Heart Rate Variability: Rescaling and Normalization

Abstract

Heart rate variability (HRV) is a useful measure to evaluate activity of the autonomic nervous system (ANS) and monitor both pathological and psychological conditions. However, HRV analysis still has difficulties with changes in HRV parameters due to an increase or decrease in the average heart rate. At present, the interpretation of the average changes in HRV datasets and their HRV parameters is not fully understood. Therefore, this study aimed to analyze how much deviation in HRV parameters occurs from rescaling tachograms and normalizing HRV datasets. Four rescaled tachograms and their corresponding normalized HRV datasets were created by increasing the average heartbeat from 50 to 110 bpm in 20 bpm steps. The difference in low frequency powers (Ln LFs) calculated between two successive rescaled groups was 0.89, 1.03, and 1.04, as the average heartbeat increased from slow to fast, while the difference in high frequency powers (Ln HFs) was 1.06, 1.53, and 1.37. However, in the four normalized HRV datasets, the difference in Ln LFs and Ln HFs between two successive normalized groups was -0.28 and -0.12, 0.31 and 0.27, and 0.31 and 0.37, respectively. The results suggest that the normalized HRV datasets are more valuable than the individual rescaled-tachogram HRV dataset for obtaining measurements using frequency-domain HRV parameters for HRV analysis in clinical applications.