Dual-Tree Complex Wavelet Packet Transform Grounded HRV Analysis for Cardiac Risk Prediction

Heart rate variability (HRV) has received a lot of attention from scientists in recent years, especially as a method of assessing physical and mental health. Due to its potential influence on autonomic nervous system (ANS) health, HRV research has received considerable attention from both conventional medicine and complementary and alternative medicine. We propose a new parameter/feature in the time domain, the standard deviation of the second-order derivative (SDSOD), to examine higher-order fluctuations in the HRV signal. Our best knowledge tells us this is the first time a second metric, turning point count (TP), often used for detecting random signals, has been utilised to evaluate an HRV signal. To analyse the HRV signal’s non-stationarity and extract its various linear and nonlinear features, we employ time–frequency representation (TFR) based approaches. We find that the linear characteristics such as SDSOD, TP, and median absolute deviation work well when comparing meditative and non-meditative data, as well as unhealthy and healthy heart data. Nonlinear features obtained from TFR analysis, such as higher order cumulants, multiscale entropies, and detrended fluctuation analysis exponents, can be utilised to complement HRV data in order to better distinguish between cardiac states. The suggested method results in a G-mean score of 0.958 and an F1-score of 0.962, with an FNR of 3.45% and an FPR of 1.32%, respectively. The next stage is to conduct tests to determine how accurately the technology can identify various critical cardiac diseases. © 2023, The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd.