Analysis of Differences in ECG Characteristics for Different Types of Drivers under Anxiety

Anxiety is a complex emotion characterized by an unpleasant feeling of tension when people anticipate a threat or negative consequence. It is regarded as a comprehensive reﬂection of human thought processes, physiological arousal, and external stimuli. The actual state of emotion can be represented objectively by human physiological signals. This study aims to analyze the diﬀerences of ECG (electrocardiogram) characteristics for various types of drivers under anxiety. We used several methods to induce drivers’ mood states (calm and anxiety) and then conducted the real and virtual driving experiments to collect driver’s ECG signal data. Physiological changes in ECG during the experiments were recorded using the PSYLAB software. The independent sample t -test analysis was conducted to determine if there are signiﬁcant diﬀerences in ECG characteristics for diﬀerent types of drivers in anxious state during driving. The results show that there are signiﬁcant diﬀerences in ECG signal characteristics of drivers by gender, age, and driving experience, in time domain, frequency domain, and waveform under anxiety. Our ﬁndings of this study contribute to the development of more intelligent and personalized driver warning system, which could improve road traﬃc safety.


Introduction
According to the statistics, more than 90% of traffic accidents are caused at least in part by human mistakes [1,2], of which many errors result from drivers' negative emotional motivations such as anxiety, anger, and contempt. Driver's emotion as a psychological response has a substantial effect on the cognitive processes, including driver's perception, judgment, action, and behavior. erefore, it is of great significance to identify driver's psychological and physiological characteristics in emotional states, in order to create safe and efficient driving.
Human emotions have a huge impact on how we live. e choices we make and the actions we take are influenced by the different types of emotions that we experience. ere have been numerous studies to investigate the complex interactions between human emotion and physiological response in social, cultural, and economic fields, including household income [3], cultural diversity [4], physical health [5,6], purchasing consumption [7,8], Internet application [9], and environmental impact [10][11][12]. For example, Jaeger et al. [10] found that anxious emotion makes people want to eat more spicy snacks and single snack intake compared to their calm state. e study by Zhang et al. [7] and Wang et al. [8] suggested that high brightness evokes people's positive emotions and low brightness evokes people's negative emotions.
In the transportation field, researchers and scholars have conducted the studies of the correlation between emotional state and driving behavior and explored the influence of human-vehicle-environment factors toward driver' mood [13][14][15]. For example, while driving, driver's anxious emotion is more likely to be induced by environmental factors such as noisy and high arousal sound [14], low road visibility [16], and driver' factors such as less driving experience [17].
Emotional states are combinations of psychological arousal and physiological response. Human emotions result in physical and physiological changes that influence behavior through autonomic nervous responses, such as electrocardiogram [18,19], respiration frequency [20,21], pulse [22], skin electricity [23,24], electromyogram [25], and skin temperature [26]. Existing research focuses on the impact of human emotion on the ECG signal properties. For example, a study by Ba et al. showed that emotion is correlated with skin resistance, heart rate, and breathing rate [27]. Takahashi et al. [28] found that the heartbeat interval becomes shorter and the ratio of low frequency band to high frequency band becomes higher in anger than in calm. Herrero-Fernández [29] found that the QT interval variability of ECG waveform is positively associated with the level of anger, and the RR interval variability of ECG waveform is negatively associated with the level of anger. e detection and warning systems for traffic safety based on drivers' ECG signals have received increasing attention. Isikli Esener [30] recognized drivers' distress level using subspace-based feature extraction on ECG signals and other physiological measurements. Balasubramanian and Bhardwaj [31] used a noncontact ECG measurement approach to determine the fatigue levels of drivers. Zhao et al. [32] measured drivers' mental fatigue according to their ECG signals. Gromer et al. [33] applied a low-cost ECG sensor to detect drivers' drowsiness, by extracting the main ECG parameters including heart rate, QRS-complex, and heart rate variability. Taherisadr et al. [34] proposed an ECG-based driver distraction detection system using convolutional neural networks.
In conclusion, there have been few attempts in the past to analyze the influence of driver's emotions on their behavioral based on physiological signals. Hence, it is essential for transportation researchers to identify driver's ECG characteristics in emotional states to gain a deep understanding of how driver's emotions affect their behavior and reactions.
is study focuses on examining the differences of ECG characteristics for various types of drivers in anxious state during driving.

Participants.
is study included 27 male drivers and 21 female drivers (age range: 22-50 y; mean age: 33.4 y). Participants were classified into three groups according to their driving propensities, which were determined by the propensity questionnaire [1]. e three groups were extraversion, middle type, and introversion, respectively. In this study, if drivers drove less than 10,000 kilometers, they would be considered as novice drivers, experienced drivers otherwise. Participants drove approximately 14,000 km miles on average. Prior to the experiment, they were told not to take any drugs that affect the brain and nervous system within one week and not to have tea, coffee, and wine that affect the mood and mental state within 48 hours. Moreover, they were asked to avoid any vigorous and high-intensity workout. Researchers provided a detailed description of the experimental design to the participants. Summary information of participants is presented in Table 1.

Emotional Induction Materials.
e materials used for emotional induction in this study were primarily obtained from the International Affective Picture System (IAPS) and the Chinese Affective Picture System (CAPS). e two databases were designed for the experimental study of emotions, by providing a set of standardized emotional stimuli according to three dimensions: pleasure, arousal, and dominance. Different types of emotion-inducing materials were applied, including audio, visual olfactory, and taste materials. Moreover, participants were also asked to carry out difficult assignment with stress, in order to induce their anxious emotion. Parts of the anxiety induction material are shown in Figure 1.

Real Driving Experiment.
e experimental route consists of a single loop, including two long sides with a length of 1.613 km (between Beijing Road and Nanjing Road) and two short sides with a length of 0.623 km (between Qingnian Road and Xincun West Road, as shown in Figure 2). e experimental equipment mainly includes two experimental vehicles, laser radar, laser ranging sensor, high-precision global positioning system, noncontact multifunction speedometer, vehicle recorder, PSYLAB human factor engineering equipment, pedal force manometer, high-definition camera, laptop, and unmanned aerial vehicle ( Figure 3). e unmanned aerial vehicle was used for recording the experimental process. Screenshots of real experimental scenes (in Xincun West Road) are illustrated in Figure 4.

Virtual Driving Experiment.
In the virtual driving experiments, a high-fidelity simulator from Japanese manufacturer FORUM 8 was used, which allowed users to construct 3D traffic environment. e Road Builder and UCwin/Road software were used in the driving simulator to build an experimentation platform of the road system with human, vehicle, and road components ( Figure 5). e driving simulator was able to collect data on interactions between drivers and vehicles under various traffic conditions. It enables researchers to collect details of useful parameters for drivers' action and vehicle performance, including distance traveled, offset from lane center, speed, acceleration, deceleration, braking, lane-changing, and Driving experience Novice (driving mileage ≤ 10,000 km) Experienced (driving mileage > 10,000 km) Novice (driving mileage ≤ 10,000 km) Experienced (driving mileage > 10,000 km) Driving tendency T 1 (extraversion); T 2 (middle type); T 3 (introversion) T 1 (extraversion); T 2 (middle type); T 3 (introversion)

Experimental Process.
e real driving experiments in anxiety were carried out during morning peak hours of 7 : 00-9: 00 and evening peak hours of 17 : 00-19 : 00 from Monday to Friday. e experimental process is shown in Figure 8.

Assessing the Level of the Induced Anxiety.
Participant's level of anxiety was detected, based on Beck Anxiety Inventory (BAI), self-perception, facial expression, and behavioral action. e BAI reflects the intensity of physical and cognitive anxiety ( Table 2). In this study, the emotion-induction experiments end when subjects obtained a score of 26 points or more. During the driving experiment, experimenters irregularly communicate to the subjects to get their emotional states. After the driving experiment, subjects were asked to watch the recorded video to report their emotional experience during driving. For more details about the process of evaluating driver's anxiety level, please refer to our another article by Guo et al. [35].

ECG Signal Data
Preprocessing. e raw ECG signals contain motion artifact, power frequency interference, and sensor internal interference noise. e PSYLAB software was used for reducing the noise in the ECG signal, as shown in Figure 9. e definitions of the parameters for denoising preprocess are given in Table 3. e comparison of ECG signal before and after denoising is shown in Figure 10. It was seen that after noise reduction, the noise can be controlled to an acceptable level. For more details about the ECG signal preprocessing, please refer to another article by Wang et al. [36].

ECG Signal Data Collection.
Each subject was involved in driving experiments multiple times. A total of 3849 groups of effective data were obtained, including 983 clusters from the real driving experiments and 2866 clusters from the driving simulators. e variables and symbols in the experiment are given in Table 4. Parts of the experimental data are given in Figure 11 and Table 5.

Driver's ECG Characteristics by Gender.
Statistical analysis was performed using SPSS Statistics 23.0 where the confidence interval was set at 95%. e independent t-test was used to determine whether there are differences in ECG indicators between female and male drivers, and the results are given in Table 6.
e results show that there are significant differences between male and female drivers in the five ECG indicators: AVHR, AVNN, RWAVE, TWAVE, and S (p < 0.05). No significant difference was found between male and female drivers in the other indicators. e results in Figure 12 and Table 6 show that female drivers have higher average heart rate and S-point peaks and lower average heartbeat, R wave peaks, and T-point peaks than male drivers. e results  Driver is asked to complete a questionnaire to show the video, audio, pictures and events that would make them feel anxious.
Driver is equipped with wireless ECG sensors. Then, soft music, colors and landscape pictures are shown to drivers.
Not anxiety state Anxiety state The driving experiment starts.The noisy audio is continuously played, and the smell and taste materials are used to stimulate the driver during driving.
The driving experiment ends (the ECG data are obtained).
The picture,video and audio (provided in the questionnaire) are shown to driver. Driver is also asked to finish difficult takes under stress. Calm state Not calm state

Advances in Civil Engineering
indicate that compared to male drivers, female drivers tend to have a faster heart rate, a shorter heartbeat interval, and a more obvious manifestation of myocardial ischemia in anxiety while driving.
Under moderate and high levels of anxiety, female drivers are more likely to experience dizziness, slow response, and fidgeting due to rapid heartbeat and poor blood flow to the heart. Moreover, females are more likely to have Note. e 21 symptoms have four levels of induction. e score of each symptom can be expressed as "1 point-none;" "2 points-mild, no major annoyance;" "3 points-moderate, feel uncomfortable but still tolerable;" "4 points-heavy, can only barely endure." e total score of 21 symptoms is 15-25 points for mild anxiety, 26-35 points for moderate anxiety, and more than 36 points for severe anxiety.   chest distress, shortness of breath, as well as discomfort in the arms, neck, and shoulders as with myocardial ischemia. ese symptoms might contribute to distraction, difficulty keeping the eyes from focusing, and slow reaction during driving. e results in Figure 12 suggest that these gender differences in the symptoms may be more pronounced in middle-aged drivers than in young ones, especially for novice and introverted drivers.  Figure 11: Distribution of driver's ECG data distribution in anxiety. (a) Driver's heart rate and frequency distribution (male * 27 y * novice * middle type). (b) Driver's heart rate and frequency distribution (female * 27 y * experienced * introversion).

Driver's ECG Characteristics by Age.
e independent t-test results (in Table 7) show that there are significant differences between middle-aged and young drivers in the seven ECG indicators AVHR, AVNN, RWAVE, TWAVE, Q, S, and UVLF/VLF (p < 0.05).
ere is no significant difference between middle-aged and young drivers in the other indicators. e results in Figure 13 and Table 7 show that young drivers have higher average heart rate, R wave peaks, and T-point peaks than middle-aged drivers. Furthermore, young drivers have lower average heartbeat interval, Q-point peaks, S-point peaks, and the ratio of ultralow frequency band to very low frequency band than middle-aged drivers. e results indicate that compared to middle-aged drivers, young drivers tend to have a faster heart rate, a shorter heartbeat interval, a higher pulse pressure, a greater sympathetic nerve activity, and a higher rate of left ventricular hypertrophy and hyperkalemia in anxiety.
In moderate and severe cases, young drivers are more likely to feel dizziness and chest distress due to rapid heartbeat and poor blood flow to the heart. Young drivers are also more likely to suffer from muscle stiffness as with hyperkalemia.
ese symptoms might contribute to slow response and maintain head-down position (vision at low location). As a result, young drivers might pay less attention on traffic environment of the sides and the straight ahead in the far while driving. ese age differences in the symptoms are more obvious in female drivers than in male ones.  Note. e significance level is 0.05. M-F, male-female. 8 Advances in Civil Engineering

Advances in Civil Engineering
Moreover, it should be noted that high levels of sympathetic nerve activity, left ventricular hypertrophy, and pulse pressure occur rarely in young individuals during driving.

Driver's ECG Characteristics by Driving Experiences.
e independent t-test results (Table 8) show that there are significant differences between novice and experienced drivers in the four ECG indicators, AVHR, AVNN, RWAVE, and S (p < 0.05). No significant difference was found between novice and experienced drivers in the other indicators. e results in Figure 14 and Table 8 show that novice drivers have higher average heart rate and R-wave peaks and lower average heartbeat interval and S-point peaks than experienced drivers. e results indicate that compared to experienced drivers, novice drivers tend to have a faster heart rate, a shorter heartbeat interval, and an aberrant ventricular conduction in anxiety.
In moderate and severe cases, novices are more likely to experience sweating and nervous intense due to rapid heartbeat. Novices are also more likely to suffer from shortness of breath as with aberrant ventricular conduction. ese symptoms might cause long fixation duration and behavioral inflexibility to react to sudden events during driving.

Conclusion
is study identified the differences of ECG characteristics for different types of drivers under anxiety. e real and virtual driving experiments were designed and conducted to collect driver ECG signal data. e data were analyzed by gender, age, and driving experience. e main findings are demonstrated as follows.
(1) Compared to male drivers, female drivers tend to have a faster heart rate, a shorter heartbeat interval, and a more obvious manifestation of myocardial ischemia in anxiety. Under moderate and high levels of anxiety, female drivers are more likely to experience dizziness, slow response, and fidgeting due to rapid heartbeat. Moreover, females are more likely to have chest distress, shortness of breath, as well as discomfort in the arms, neck, and shoulders as with myocardial ischemia. (2) Compared to middle-aged drivers, young drivers tend to have a faster heart rate, a shorter heartbeat interval, a higher pulse pressure, a greater sympathetic nerve activity, and a higher rate of left ventricular hypertrophy and hyperkalemia in anxiety. In moderate and severe cases, young drivers are more likely to feel dizziness and chest distress due to rapid heartbeat. Young drivers are also more likely to suffer from muscle stiffness as with hyperkalemia. (3) Compared to experienced drivers, novice drivers tend to have a faster heart rate, a shorter heartbeat interval, and an aberrant ventricular conduction in anxiety. In moderate and severe cases, novices are more likely to experience sweating and nervous intense due to rapid heartbeat. Novices are also more likely to suffer from shortness of breath as with aberrant ventricular conduction.
Our findings of this study suggest that ECG signals closely reflect driver's emotional state and can be used to detect driver's physical state. e findings also contribute to the development of the intelligent and personalized driver warning system, which could improve road traffic safety. Further studies are required to gather additional ECG data for different types of drivers and determine the factors affecting the ECG characteristics in emotional states.

Data Availability
e data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
e authors declare that there are no conflicts of interest.