Drowsiness detection using heart rate variability

Abstract

It is estimated that 10-30 % of road fatalities are related to drowsy driving. Driver's drowsiness detection based on biological and vehicle signals is being studied in preventive car safety. Autonomous nervous system activity, which can be measured noninvasively from the heart rate variability (HRV) signal obtained from surface electrocardiogram, presents alterations during stress, extreme fatigue and drowsiness episodes. We hypothesized that these alterations manifest on HRV and thus could be used to detect driver's drowsiness. We analyzed three driving databases in which drivers presented different sleep-deprivation levels, and in which each driving minute was annotated as drowsy or awake. We developed two different drowsiness detectors based on HRV. While the drowsiness episodes detector assessed each minute of driving as "awake" or "drowsy" with seven HRV derived features (positive predictive value 0.96, sensitivity 0.59, specificity 0.98 on 3475 min of driving), the sleep-deprivation detector discerned if a driver was suitable for driving or not, at driving onset, as function of his sleep-deprivation state. Sleep-deprivation state was estimated from the first three minutes of driving using only one HRV feature (positive predictive value 0.80, sensitivity 0.62, specificity 0.88 on 30 drivers). Incorporating drowsiness assessment based on HRV signal may add significant improvements to existing car safety systems.

Sleep-wake classification via quantifying heart rate variability by convolutional neural network

“…These physiological facts indicate that we can distinguish between the wake and sleep stages by taking heart rate into account. There have been several works in this direction, such as [25,31,28,49,2,51,47].…”

Sleep Deprivation in Young and Healthy Subjects Is More Sensitively Identified by Higher Frequencies of Electrodermal Activity than by Skin Conductance Level Evaluated in the Time Domain

“…Despite the spectral overlap, HRVHF is widely used as a marker of parasympathetic function and HRVLF is used as a marker of sympathetic function (Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology, 1996). Many studies have used indices of HRV to track the effects of sleep deprivation on humans (Nakano et al, 2000; Zhong et al, 2005; Viola et al, 2008; Pagani et al, 2009; Fogt et al, 2010, 2011; Chua et al, 2012; Glos et al, 2014; Vicente et al, 2016), with different levels of success. These studies suggest that HRV has the potential to track several effects of sleep deprivation.…”

Feature Extraction and Evaluation for Driver Drowsiness Detection Based on Thermoregulation

“…In this study, two-minute-long drowsy and alert events were classified based on the ratio of low (0.04-0.15 Hz) to high frequency (0.15-0.40 Hz) variability in the heart rate obtained from photoplethysmography. In another study, ECG measurements were used to detect drowsiness based on HRV as well as respiratory frequency [32]. Here, a positive predictive value (PPV), sensitivity, and specificity of 96%, 59%, and 98%, respectively, was reported.…”