Feasibility study on driver's stress detection from differential skin temperature measurement

Abstract

Prolonged monotonous driving may lower a driver's awareness level as well as increasing their stress level due to the compulsion to maintain safe driving, which may result in an increased risk of a traffic accident. There is therefore an opportunity for technological assessment of driver physiological status to be applied in-car, hopefully reducing the incidence of potentially dangerous situations. As part of our long-term aim to develop such a system, we describe here the investigation of differential skin temperature measurement as a possible marker of a driver's stress level. In this study, healthy male (n=18) & female (n=7) subjects were investigated under environment-controlled conditions, whilst being subjected to simulated monotonous travel at constant speed on a test-course. We acquired physiological variables, including facial skin temperature which consists of truncal and peripheral skin temperatures (Ts) using thermography, beat-by-beat blood pressure (BP), cardiac output (CO), total peripheral resistance (TPR), and normalized pulse volume (NPV) used as an indicator of local peripheral vascular tone. We then investigated the driver's reactivity in terms of skin temperatures with this background of cardiovascular haemodynamics. We found that the simulated monotonous driving produced a gradual drop in peripheral Ts following the driving stress, which, through interpretation of the TPR and NPV recordings, could be explained by peripheral sympathetic activation. On the other hand, the truncal Ts was not influenced by the stress. These findings lead us to suggest that truncal-peripheral differential Ts could be used as a possible index indicative of the driver's stress.

A Critical Review of Proactive Detection of Driver Stress Levels Based on Multimodal Measurements

“…As a driver's stress and mental workload levels increase, HF values decrease while LF values increase. In addition to HR and HRV signals, a driver's beat-by-beat mean blood pressure and normalized blood volume pulse have been shown to have a correlation with the driver's stress level (Yamakoshi et al 2008). Total peripheral resistance is also shown as a cardiovascular feature related to driver stress (Yamakoshi et al 2008).…”

“…Nose, cheek, forehead, and jaw regions on the face, which contain abundant arterioles and arteriovenous anastomoses, are used as skin temperature features to monitor drivers' stress and mental workload (Yamakoshi et al 2008). Yamakoshi et al (2008) measured drivers' skin temperature, cardiovascular response, and contextual parameters to stressful driving situations. Temperature data through the finger (Tsf), nose (Tsn), cheek (Tsch), jaw (Tsj), and forehead regions (Tsfh) are collected.…”

“…,Rigas et al (2008),Yamakoshi et al (2008),Akbas (2011),,Deng et al (2013),Wang et al (2013),,Munla et al (2015),Rodrigues et al (2015),Lanatà et al (2015), andHeikoop et al (2017) Ü Heart rate (HR): mean, standard deviation (SD) Ü PPG: mean, variance, SD, signal energy, time duration, bandwidth, time-bandwidth product dimensionality, cardiac output, PPG instantaneous, pulse height, PPG rise, PPG fall, PPG cardiac, NPV Ü Heart rate variability (HRV) (time domain): mean of the first difference (MFD), mean of R-to-R intervals (MRR), SDRR, MNN, SDNN, RMSSD, pNN20, pNN50, NN50, triangular interpolation of the NN interval histogram (TINN), heart rate variation from baseline Ü HRV (frequency domain): LF/HF ratio, HF, LF, very low frequency (VLF), total power(TP) Ü HRV (nonlinear analysis): sample entropy, detrended fluctuation analysis (DFA), DFA-α 1 and DFA-α 2, SD1, SD2 Ü HRV (time-frequency domain analysis): VLF, LF, and HF by using wavelet and short time Fourier transform (STFT) analysis EDA Avcı Akbaş and Yüksel (n.d.), Healey and Picard (2000), Healey and Picard (2005), Rigas et al (2008), Katsis et al (2008), Akbas (2011), Rigas et al (2011, 2012), Deng et al (2013), Singh and Queyam (2013b), Singh et al (2013), Lanatà et al (2015), and Ollander et al (2017) Ü Mean, variance, SD, mean amplitude, peak rise time, peak amplitude, half-recovery, peak energy, rise rate average, decay rate average, percentage decay, number of peaks, rate of SCR, first absolute difference (FAD), MFD, mean rise duration, number of SCR, latitude, the amplitude sum of SCRs (AmpSum), the phasic area under curve, the max phasic amplitude, mean of SCR, variance of SCR, mean of SCL, variance of SCL, number of the nonspecific response, frequency, magnitude, duration, area of orienting responses, sum of frequency of occurrence (SF), sum of durations, sum of magnitudes (SM), sum of the estimated areas (SA), mean of positive derivative, mean of absolute derivative, proportion of positive samples in derivative, number of local maxima Respiratory Response Healey and Picard (2000), Katsis et al (2008), Katsis et al (2011), Rigas et al (2011), Akbas (2011), Deng et al (2012), Rigas et al (2012), Lanatà et al (2015), and Ollander et al (2017) Ü Time Domain: mean of RESP, SD of RESP, average number of contractions/minute, RSPR, HR/RESP ratio, dominant respiration frequency, PPG respiratory rate (PPGRSP); max, min, and mean breathing amplitude; standard error of the mean (SEM), difference between max and min value of breathing amplitude, skewness, kurtosis, entropy of the resulting spectrum, MFD, mean of the second difference (MSD); standard deviation of the first difference (SDFD), standard deviation of the second difference (SDSD); second difference breathing amplitude (SDBA), maximal respiration-mean of respiration (range) Ü Frequency Domain: RESP frequency (0-0.1), RESP frequency (0.1-0.2), RESP frequency (0.2-0.3), RESP frequency (0.3-0.4) EMG Katsis et al (2008), Soman et al (2014), and Zheng et al (2015) Ü Mean, RMS, ZCR, RMQ ST Yamakoshi et al (2007), Shastri et al (2008), and Yamakoshi et al (2008) Ü Finger temperature (Tsf), nose temperature (Tsn), cheek temperature (Tsch), jaw temperature (Tsj), forehead temperature (Tsfh) Blood Pressure Yamakoshi et al (2008) Ü Systolic blood pressure, mean blood pressure, diastolic blood pre...…”

Mobile Stress Recognition and Relaxation Support with SmartCoping: User-Adaptive Interpretation of Physiological Stress Parameters

“…However, skin conductance also increases as a result of regulating body temperature, which diminishes its reliability as an indicator unless it is measured in controlled settings. Facial skin temperature is another indicator for stress [45], as is blood pressure.…”

“…Another study uses differential skin temperature, measured e.g. at the periphery with sensors and at the face using an infrared camera [45] .…”