Options
A Comparative Study of Heart Rate Variability Methods for Stress Detection
Date Issued
01-01-2022
Author(s)
Abstract
Heart Rate Variability (HRV) is a widely accepted technique used to measure the stress level of individuals. The state of the art HRV features namely Hilbert spectral and Detrended Fluctuation Analysis (DFA) estimates have opened ways to measure mental state of the individual. The HRV spectral estimates Instantaneous Amplitude of Low Frequency band (LFiA) & Instantaneous Amplitude of High Frequency band (H FiA) derived from Hilbert Transform provides better categorization of mental stress states than conventional frequency parameters due to joint 2-D representation of the low and High frequency HRV bands. Another HRV based approach DFA, which is robust against non-linearity and non-stationarity of cardiac time series caused by complex interactions, helps in providing reliable HRV interpretation. Based on the research evidences for both Hilbert and DFA estimates, it was observed that the use of Hilbert spectral estimates in stress assessment was not validated under free living condition and the application of DFA for mental stress assessment of individuals was not studied. In this work the utility of Hilbert Transform and DFA in HRV based stress assessment was studied under two different settings (confined and free living). The first objective was to determine whether DFA can be used to delineate between two mental states (baseline and stress), under both confined and free living conditions, and to quantify its discriminatory power in the context of mental stress detection. The second objective was to examine the utility of Hilbert estimates in determining mental state under free living conditions. The third objective was to compare the discriminatory power of DFA and Hilbert Transform in stress state detection. From this study, it was observed that both Hilbert and DFA methods can be used to delineate between two mental states under both confined and free living conditions. From the comparative analysis, it was observed that Hilbert estimates showed better discriminatory power than DFA under both the settings.