Content uploaded by John Yuan
Author content
All content in this area was uploaded by John Yuan on Jun 14, 2014
Content may be subject to copyright.
Accepted Manuscript
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our authors and readers we are providing this early version of the manuscript. The manuscript will
undergo copyediting, typesetting, and review of the resulting proof before it is published in its final
form. Please note that during the production process errors may be discovered which could affect the
content, and all legal disclaimers that apply to the journal pertain.
.
CHANGES IN PHYSIOLOGICAL PARAMETERS
INDUCED BY SIMULATED DRIVING TASKS:
MORNING VS. AFTERNOON SESSION
Wen-Chieh Liang, John Yuan, D. C. Sun, and
Ming-Han Lin
Reference: 2007028
To appear in: Journal of the Chinese Institute of
Industrial Engineers
Received date: April 2007
Revised date: May 2007
Accepted date: April 2008
CHANGES IN PHYSIOLOGICAL PARAMETERS INDUCED
BY SIMULATED DRIVING TASKS: MORNING VS.
AFTERNOON SESSION
Wen-Chieh Liang* and John Yuan
Industrial Engineering and Engineering Management, National Tsing Hua University
101, Section 2, Kuang-Fu Road, Hsinchu City, Taiwan 300, R.O.C.
D. C. Sun
General Manager, Taiwan Scientific Corporation
Ming-Han Lin
Automation Engineering, Ta Hwa Institute of Technology
ABSTRACT
Introduction: Driving fatigue is one of the most common causes for traffic accidents. Im-
mobilization of legs, hip, and waist is thought to play a major role in driving fatigue, as it
hinders blood circulation and induce hemodynamic changes. Objective: the objective of the
study was to monitor changes in physiological parameters before and after indoor simulated
driving tasks conducted in the morning as well as afternoon sessions. Method: 40 young
male subjects were randomly divided into morning (group A) and afternoon (group B) ses-
sions and participated in the 90-min simulated in-door driving task. Before and after the task,
BP (blood pressure), HR (heart rate), and HRV (heart rate variability) parameters were
measured using a novel wrist monitor ANSWatch which utilized built-in bio-sensors in the
cuff to acquire radial pulse waves directly. Palm temperatures were measured by a
high-precision thermometer. A questionnaire ranking driving fatigue was filled by each
volunteer before and after the driving task. Results: (1) from paired T-tests, both the
morning and afternoon driving tasks caused decreases in HR and palm temperatures, and
increases in HRV and VLF(AU) (Very Low Frequency(Absolute Unit)); For the morning
session, LF(AU) (Low Frequency(Absolute Unit)) and LF(NU)(Low Fre-
quency(Normalized Unit)) increased while HF(NU) (High Frequency(Normalized Unit))
decreased; In contrast, LF(NU) and LF/HF decreased while HF(NU) increased for the af-
ternoon session (all changes p<0.05). Systolic pressure was maintained in the morning ses-
sion but dropped in the afternoon session (p<0.05). (2) From One-way and Two-way
MANOVA analyses, there was no significant difference between morning and afternoon
session for the entire group of physiological parameters measured before or after driving
tasks; However, LF(AU), LF(NU), and LF/HF three individual parameters measured before
driving were higher in the afternoon session than in the morning session (p<0.05). (3) From
written questionnaire, all subjects felt some degree of fatigue following the driving task. No
statistical difference existed between the two driving sessions in terms of fatigue score
baseline or score change due to driving. Conclusion: Multiple physiological parameters
showed significant changes after simulated driving tasks. Distinct trends were found be-
tween the two driving sessions. In the morning session, poor circulation in the lower body
(limbs, abdomen, and hip) caused decrease in palm temperatures and heart rate, but blood
pressures were maintained due to activation of the sympathetic nervous system as evidenced
by increased HRV, LF(AU), and LF(NU). For the afternoon session, palm temperatures,
heart rate, and systolic pressure were all lowered. Para-sympathetic nervous system was ac-
tivated (indicated by increased HF(NU)) prompting the body to enter a sleepy state, which
greatly increases accident risks in actual road driving. Monitoring of multiple physiological
parameters in the study had gained great insight into mechanisms of homeostasis and pro-
vided a foundation in the future work to quantify driving fatigue in terms of degree of de-
viation from homeostatic states.
Keywords: Driving Fatigue, Heart Rate Variability (HRV), Homeostasis, Autonomic
* Corresponding author: richard168@mail.e88.com.tw
Nervous System, Blood Pressure, Simulated Driving, ANSWatch
1. INTRODUCTION
While driving fatigue is still a vaguely defined
term physiologically, its effect on traffic accidents is
well documented. Numerous statistics and studies
have shown that long hour driving resulted in physical
tiredness and slowdown in mental judgment. In the
report published by Shinar in 1978 [18], a significant
portion of highway accidents were attributed to driv-
ing fatigue. National Transportation Safety Board in
U.S. investigated 286 accidents involving commercial
vehicles and discovered that 38% of accidents were
caused by the drivers’ drowsiness or neglect (Harris
and Mackie, 1972). Overall, driving fatigue remains
one of the most probable causes for traffic accidents. A
better understanding in driving fatigue on the physio-
logical level will lead to new development for effec-
tive prevention.
In one aspect of driving, drivers often fail to
maintain fresh concentration on the road with repeti-
tious and unexciting scenery. In another aspect of
driving, the limited waist, hip and leg spaces constrain
the lower body (limbs, abdomen, and hip) from active
movement. The “pumping” action by the leg muscle
contraction, which forces the venous blood back to the
heart (such as during walking or exercise), is largely
lost. It is our belief that immobilization of the lower
body plays a major role in driving fatigue, as it hinders
systemic blood circulation and induces significant
hemodynamic changes. This belief is consistent with
widely discussed mechanisms in driving fatigue, in-
cluding hypoxia in brain, blood pressure drop, and
below-normal heart rate.
It is well known that when local or systemic cir-
culation is obstructed, ANS(Autonomic Nervous Sys-
tem) in the body is activated swiftly. Through its
sympathetic and parasympathetic branches, ANS
helps the cardiovascular system to maintain proper
blood supply under these compromised circumstances.
If the activation and execution of ANS is effective,
deviations of physiological parameters from the ho-
meostatic states can be avoided or minimized. On the
other hand, any significant change in vital physio-
logical parameters from baseline (such as body tem-
perature or blood pressures) may point to an “ex-
hausting” body which is unable to respond to
physiological needs.
HRV (Heart rate variability) has been used in
various studies for the assessment of physiological
states. The subconscious cyclic variation in heart rate
period is commonly analyzed in both time and fre-
quency domains to give rise to parameters that are
linked to total ANS activity (HRV or SDNN, TP or
total power), sympathetic activity (LF (AU) and LF
(NU)), parasympathetic (or vagal) activity (HF (AU)
and HF (NU)), and sympatho-vagal balance (LF/HF)
indexes. Hjortskov et al. [7] and Garde et al. [5] both
monitored HRV parameter changes in volunteers
before and after a computer task during which various
degrees of mental stresses were introduced. Hjortskov
et al. [7] indicated that stressors led to changes in HRV
(increase in LF(AU), HF(AU), and LF/HF compared
to those under resting conditions), and a sustained
increase in blood pressures (SYS and DIA). Garde et
al. [5] also reported an increase in heart rate, blood
pressure, and LF (NU), and a decrease in TP(AU) and
HF (NU) in response to a physically demanding ref-
erence computer task. Wahlstrom et al. [26] also in-
troduced time and verbal stresses during a mouse -
driven computer task to investigate the physiological
and psychological changes based upon heart rate,
blood pressures (SYS and DIA), and HRV. Increases
in both the physiological (HR, BP, LF/HF) and psy-
chological reactions were observed compared to con-
trol conditions. These reports suggest that physical and
mental stresses may cause the activation of sympa-
thetic nervous system as indicated by increased BP,
HR, LF, and LF/HF.
In a similar manner, several authors investigated
the effect of simulated flight on physiological pa-
rameters [9][11][25]. Their general finding is that the
complexity of a pilot’s task in operating a flight often
caused an increase in HR and BP (SYS and DIA), and
a decrease in HRV. Lee [11] clearly showed that when
the pilots conducted tasks that required high concen-
tration, such as during take-off and landing, their heart
rates increased significantly. Among various tasks
performed by pilots (take off, climb and cruise, de-
scent and approach, and landing), HRV was seen
lowest during approach as it was the most critical
period of piloting.
In the area of indoor simulated driving tests
[8][14][15][16][17][27], Yang et al. [27] utilized ECG
to monitor the drivers’ HRV changes. They discovered
four HRV parameters that were significantly changed
after driving, namely increased HRV (or SDNN),
increased LF (AU), decreased HF (NU), and increased
LF/HF. Yang et al. [27] also reported that as the degree
of fatigue increased (indicated by increasing driving
hour), SDNN (equivalent to HRV in this study),
LF(NU) and LF/HF all increased while HF(NU) de-
creased progressively. The authors believed that the
increase in LF/HF was an indication of increase in
degree of driving fatigue, as the balance of ANS
shifted towards the sympathetic branch. Li et al. (2003)
[16] also based their HRV study on ECG and found
three HRV parameters with significant changes after
simulated driving, including increased LF (NU), de-
creased HF (NU), and increased LF/HF. After three
hours of continuous driving, the drivers showed an
increase in response time, a drop in judgment accuracy,
and a lower heart rate. Subjective written question-
naire also showed increased symptoms of driving
fatigue. They proposed using HRV as a quantitative
index of driving fatigue. Li et al. [14,15,17] further
studied the effect of acupuncture on driving fatigue.
Their findings suggested that driving fatigue induced
physiological changes could be attenuated by acu-
puncture. Actual road driving involves much higher
mental and physical loads than the indoor simulation.
Both Fumio et al. [4] and Miloševic et al. [21] ob-
served an increase in BP on taxi drivers on long duty
schedules. It is to be noted that the levels of mental and
physical workloads, time duration, as well as the de-
gree of body immobilization are all different among
the above were cited studies as well as in this research.
We thus expected to see unique body responses in the
volunteers during the 90-min indoor driving employed
in this study.
In this study, we monitored multiple physio-
logical parameters, including palm temperatures (left
and right), HR, BP (SYS and DIA), and HRV pa-
rameters before and after a 90-min indoor driving task.
In addition, a written questionnaire was filled by each
participant before and after the driving task to gauge
the subjective evaluation of driving fatigue. The study
was divided into two sub-groups, one conducted in the
morning, and the other in the afternoon, so that body
reactions to long-hour driving at different times of the
day could be investigated. It is our hope that the results
of this first phase in a series of studies could provide
useful information for the definition of driving fatigue
based upon physiological parameter changes.
2. METHODS
In-door simulated driving (instead of road driv-
ing) was selected in consideration of cost, safety, and
control of variables.
2.1 Subjects
All volunteers gave their informed consent before
the study. To avoid the influences of gender and age
on HRV [13][28], a total of 40 male subjects in the age
of 23.3±1.9 years old (Table 1; all college students or
graduates) were selected to take the driving test. All
subjects were currently healthy and without any
medical treatments. They were instructed to have
sufficient sleep in the previous night and not to eat,
drink, or exercise one hour prior to the test. All sub-
jects were confirmed to be in fresh or non-fatigue
conditions before driving when reporting to the labo-
ratory.
Table 1. Characteristics of subject
Item Average
Age 23.3±1.9 (years-old)
Height 169.2±6.2 (cm)
Weight 70.6±9.6 (kg)
Body Fat Index* 22.5±6.1 (%)
* Body fat index measured by a commercial instrument
manufactured by TANITA corporation (Model: ULT
2000 / ULT 2001, Tokyo, Japan)
2.2 Variables
In the experimental design, gender, age and am-
bient temperature (and humidity) were controlled. The
independent variables (effects) were driving session
(morning or afternoon) and driving task (before or
after). The dependent variables were physiological
parameters which included blood pressures [systolic
(SYS) and diastolic (DIA)], heart rate (HR), heart rate
variability (HRV), sympathetic nerve activity indexes
[LF(AU) and LF(NU)], parasympathetic nerve activ-
ity indexes [HF(AU) and HF(NU)], sympatho-vagal
balance index (LF/HF), and temperature of left and
right palms.
2.3 In-door simulated driving
The test room was temperature controlled at
22±2 .℃ A simulate highway scenery was projected
onto a 178 (cm) x 178 (cm) white screen using a
computer and a projector (Figure 1). The driver’s seat
was about 3-4 m away from the screen. There were
trees on the left and walls on the right side of the
four-lane, two-way highway. The driver must operate
the wheel to keep the vehicle on the designated lane
without hitting trees or walls. A red warning scale
appeared on the left side of the windshield which
would expand in area vertically if the vehicle location
deviated from the designated lane. Laud sounds would
go off if the vehicle got too close to or made contact
with road trees or walls. The driving task lasted 90
minutes. While the driver’s main task was to operate
the wheel, he was not required to step the gas or the
break peddle (both not equipped). Instead, a constant
driving speed (the road view) was provided by the
computer. Such a simulation is close to a highway
driving where the traffic conditions are more constant
as compared to a local in-town driving where the
driver must change the speed frequently using gas and
break peddles.
Figure 1. Driving simulator
2.4 Apparatus and materials
Experimental apparatus consisted of a HP note-
book computer, a computer projector, a driving wheel,
a timer watch, a body weight and fat balance, a high
precision thermometer (+0.1C), and ANSWatch.
Software included highway scenery simulator, Win-
dows XP, SPSS 12.0, and “ANSWatch Manager Pro”
data analyzer.
2.5 Experimental procedures
40 subjects were randomly divided into two
groups (A&B). Group A conducted driving tests in the
morning (8:30~11:00AM) while Group B in the af-
ternoon (2:00~4:30PM). When reported to the test
room, each volunteer took a 20-min rest first and then
underwent thermometer (both palms) and ANSWatch
tests. The two tests needed about 7 minutes. Data in
ANSWatch was downloaded to a notebook com-
puter immediately following the test for review and
storage. The driving task followed which lasted for 90
minutes. After driving, each volunteer was tested
again for palm temperatures and ANSWatch. In
addition, a written questionnaire consisting of 14
questions related to fatigue (similar to those developed
by authors in [17]) was filled by the volunteers before
and after driving. The entire testing program is illus-
trated in Figure 2. The list of questions is shown in
Table 2, while the quantitative scale for each question
is shown in Table 3. A score of 4 or higher is an in-
dication of positive response to the question.
20-min 5-min 9-min 90-min 9-min 5-min
Rests Ques. Tests Driving task Tests Ques.
138 min
Figure 2. Experimental procedures time chart
Table 2. Questionnaire for feeling of driving fatigue
No Symptom
1 Body tiredness
2 Loss of concentration
3 Desire to lie down
4 Anxiety
5 Lack of energy
6 Mental response slowdown
7 Headache
8 Shoulder stiffening
9 Waist pain
10 Lower body numbness
11 Eye fatigue
12 Feeling of sleepiness
13 Feeling of vomit
14 Hand and foot trembling
Table 3. Quantitative scale (1-7)
Scale Fatigue description
1 No such feeling
2 Negligible feeling
3 Some feeling
4 Clear feeling
5 Strong feeling
6 Very strong feeling
7 Extremely strong feeling
2.6 Physiological parameters analyses
During the 6-min test, ANSWatch (Figure 3)
first used the oscillatory method to obtain heart rate,
systolic pressure, and diastolic pressure. It then con-
ducted a standard 5-min HRV test. The
piezo-electrical sensors in the cuff picked up blood
pressure waveforms produced by the radial artery,
with the aid of an air pouch pressure controlled by an
air pump and a release valve. Peak-to-peak intervals
were determined followed by time and frequency
domain analyses. The HRV analysis followed closely
the 1996 international standard [24], consisting of the
following steps:
(1). The original data was fed through a low pass
FIR filter at 0 to 14 Hz.
(2). Fundamental frequency was determined based
upon the first 5-second data.
(3). The primary peak in each cycle was determined.
(4). Peak-to-peak intervals were calculated.
(5). Time-domain HRV parameters (mean period or
heart rate; variance and standard deviation of
peak-to-peak intervals) were calculated. Peak
intervals greater than 4*standard deviation were
removed and not replaced.
(6). Peak-to-peak intervals were re-sampled to 1024
points with interpolation and Hamming window
adjustment.
(7). Fast Fourier Transform (FFT) was performed
with Hamming window adjustment.
(8). Integrations of power spectral density between
0.0001 and 0.04 Hz for the very low frequency
component (VLF), between 0.04 and 0.15 Hz
for the low frequency component (LF), and
between 0.15 and 0.4 Hz for the high frequency
component (HF) respectively were conducted.
(9). Frequency-domain HRV parameters {VLF
(AU), LF (AU), HF (AU), LF (NU) [equal to
LF/(LF+HF)*100], and HF (NU) [equal to
HF/(LF+HF)*100]} were calculated.
Figure 3. ANSWatch wrist monitor
It is noted above that irregular heartbeats (defined
as those with peak intervals greater than 4 standard
deviations in the 5-min test data, as caused by cardiac
arrhythmia or body movement) were excluded from
the raw data prior to HRV analysis (as recommended
by the 1996 Standard [24]). For clarity, the HRV pa-
rameters used in the study are listed below with asso-
ciated physiological meanings and units:
(1). HR：Heart rate (beat/min)
(2). HRV：Total ANS activity index (ms); equal to
standard deviation of adjacent peak-to-peak in-
tervals SDNN defined in 1996 standard.
(3). VLF(AU) ：Very Low Frequency (Absolute
Unit) (frequency range 0.0001~0.04 Hz) (ms2);
its physiological meaning not defined by 1996
Standard
(4). LF(AU) ：Low Frequency (Absolute Unit)
(frequency range 0.04 ~0.15 Hz) (ms2); sym-
pathetic (and some parasympathetic) nervous
activity index
(5). HF(AU) ：High Frequency (Absolute Unit)
(frequency range 0.15~0.4 Hz) (ms2); para-
sympathetic nervous activity index
(6). LF(NU)(%) ：Low Frequency(Normal Unit),
[LF/(TP-VLF)]*100; contribution of sympa-
thetic nervous activity
(7). HF(NU)(%) ：High Frequency(Normal Unit)
[HF/(TP-VLF)]*100; contribution of parasym-
pathetic nervous activity
(8). LF/HF ：Ratio of LF(AU) to HF(AU);
sympatho-vagal balance index
Although the physiological meaning for VLF
(AU) was reported by ANSWatch, it is not defined in
the 1996 Standard. The authors decided to report VLF
(AU) data to aid discussions.
2.7 Data collection
Up to date, most HRV studies have been using
ECG due to its availability in research laboratories. A
few studies have based their HRV measurements on
finger blood pressure waveforms using an optical
sensor [1][6][20]. They reported data accuracy in
terms of correlation coefficient in the range of 0.75 to
0.99 when compared to ECG. In this paper, we are
introducing a new wrist monitor ANSWatch (Tai-
wan Scientific Corporation, Taipei, Taiwan; Taiwan
DOH (Department of Health) Approval number
001525) which employs multiple piezo-electrical
sensors enclosed in the cuff to directly measure the
blood pressure waveforms in the radial artery. Ac-
cording to the company documents submitted to Tai-
wan Department of Health and published literature
[12][22][23], the device accuracy (correlation coeffi-
cient) is in the range of 0.90 to 1.0 using ECG as the
control. This portable device requires neither elec-
trodes nor other disposables, and can conduct tests in
sitting or lying postures. Each ANSWatch test takes
about 6-minutes and outputs eight patient parameters
on the LCD screen (heart rate HR, systolic pressure
SYS, diastolic pressure DIA, heart rate variability
HRV (or standard deviation of 5-min peak-to-peak
intervals SDNN), low frequency (normalized) LF
(NU), high frequency (normalized) HF(NU), sym-
patho-parasympathetic balance index LF/HF, and
number of irregular heartbeats (cardiac arrhythmia).
Upon data download to a PC and using the accompa-
nied software (ANSWatch Manager Pro), more
HRV parameters can be calculated (such as low fre-
quency (absolute) LF (AU), high frequency (absolute)
HF (AU), total power TP, very low frequency (abso-
lute) VLF (AU), and square root of the mean of the
sum of the squares of differences between adjacent
peak intervals RMMSD etc.)
2.8 Statistical analyses
Student’s T tests (two-tailed) were used
throughout the entire study to determine the signifi-
cance of parameter changes before and after the driv-
ing task for respective driving sessions. Furthermore,
One-way and Two-way MANOVA (Multivariate
analysis of variance) analyses were used to examine
any group difference or interactions. The question-
naire results were analyzed using the same methods.
3. RESULTS
3.1 Variation in physiological parameters
before and after driving (pair T test
analyses)
Table 4 and Table 5 show the test results for the
morning session (Group A) and the afternoon session
(Group B) respectively.
Table 4. Physiological parameters before and after
driving task for the morning session
Parameters Before driving After driving t-value DF ap-value
SYS 113.8±9.0 113.8±9.2 0.02 19 0.987
DIA 73.4±1.8 73.6±1.9 0.44 19 0.666
HR 70.2±11.3 65.8±8.6 -3.09 19 0.006
HRV 44.5±14.7 58.7±16.4 6.20 19 0.000
LF (AU) 468.4±302.7 716.7±434.5 3.12 19 0.006
LF (NU) 46.3±17.1 54.3±14.5 2.26 19 0.036
HF (AU) 580.0±518.1 591.5±383.4 0.11 19 0.916
HF (NU) 53.7±17.1 45.6±14.5 -2.26 19 0.036
VLF (AU) 1142.4±983.8 2366.2±1584.3 5.62 19 0.000
LF/HF 1.1±0.8 1.3±0.7 1.45 19 0.164
TLP 36.5±0.6 35.4±1.7 -3.01 19 0.007
TRP 36.6±0.7 35.6±1.7 -2.65 19 0.016
a: DF (degree of freedom)
From Table 4, HR, HRV, HF(NU), LF(AU) &
LF(NU), VLF(AU), left palm temperature (TLP) and
right palm temperature (TRP) exhibited significant
changes (p<0.05) after the driving task conducted in
the morning for Group A, while changes in SYS, DIA,
HF (AU) or LF/HF did not reach statistical signifi-
cance.
Table 5. Physiological parameters before and after
driving task for the afternoon session
Parameters Before driving After driving t-value DF ap-value
SYS 118.7±7.7 111.4±5.0 -4.16 19 0.001
DIA 73.9±2.3 73.9±2.0 0.11 19 0.917
HR 71.4±9.2 67.0±9.7 -3.33 19 0.003
HRV 48.7±16.9 56.3±17.7 2.71 19 0.014
LF (AU) 879.8±847.6 817.9±545.1 -0.39 19 0.700
LF (NU) 63.2±14.8 53.5±16.7 -3.21 19 0.005
HF (AU) 510.9±465.7 742.4±565.3 3.02 19 0.007
HF (NU) 36.8±14.8 46.6±16.7 3.21 19 0.005
VLF (AU) 1253.7±823.9 1907.1±1372.8 2.70 19 0.014
LF/HF 2.1±1.2 1.5±1.0 -2.69 19 0.015
TLP 36.7±0.8 35.7±1.8 -3.78 19 0.001
TRP 36.8±0.8 36.0±1.9 -2.87 19 0.010
a: DF (degree of freedom)
From Table 5, SYS, HR, HRV, LF(NU),
HF(AU), HF(NU), VLF(AU), LF/HF, left palm tem-
perature(TLP) and right palm temperature (TRP) all
exhibited significant changes (p<0.05) after the driv-
ing task conducted in the afternoon for Group B, while
changes in DIA and LF(AU) did not reach statistical
significance. It is noted that average systolic pressure
went down from 118.7 (before driving) to 111.4
mmHg (after driving) with a p-value of 0.001.
3.2 Variation in physiological parameters
(Multivariate Analysis of Variance,
MANOVA)
3.2.1 One-way MANOVA (morning vs. after-
noon)
(1). Before driving
a. Entire group of physiological parameters
This analysis was based upon the entire group of
physiological parameters measured before driving in
the study (a total of 12, see Table 4 or 5) to examine
any statistical difference between the morning and the
afternoon session. The results are shown in Table 6.
Table 6. Comparison between sessions for the entire
group of physiological parameters before driving
Effect Test Value F Hypothesis DF
a Error DF
ap-value
Driving
session
Wilks'
λ 0.673 1.23 11 28 0.309
a: DF (degree of freedom)
From Table 6, there was no significant group
difference (p=0.309) between the morning and the
afternoon session for the 12 physiological parameters
taken before driving.
b. Individual physiological parameters (inde-
pendent T-test)
Individual physiological parameters were exam-
ined by independent T-tests for any group difference
between the morning and the afternoon session. Re-
sults are shown in Table 7.
Table 7. Comparison between sessions for individual
physiological parameters before driving
Session average
Parameters Morning Afternoon
Difference
b between
session
Stan-
dard
devia-
tion
DF ap-value
SYS 113.80 118.65 4.85 2.65 38 0.075
DIA 73.45 73.90 0.45 0.65 38 0.498
HR 70.20 71.40 1.20 3.27 38 0.716
HRV 44.55 48.70 4.15 5.02 38 0.414
HF (AU) 580.00 510.85 -69.15 155.77 38 0.660
HF (NU) 53.70 36.80 -16.90 5.06 38 0.002
LF (AU) 468.45 879.75 411.30 201.26 38 0.048
LF (NU) 46.30 63.20 16.90 5.06 38 0.002
VLF (AU) 1142.40 1253.70 111.30 286.95 38 0.700
LF/HF 1.11 2.13 1.02 0.32 38 0.003
TLP 36.54 36.71 0.16 0.22 38 0.474
TRP 36.61 36.78 0.17 0.25 38 0.507
a: DF (degree of freedom)
b: Difference was defined as value (afternoon ses-
sion) – value (morning session)
From Table 7, LF(AU), LF(NU) and LF/HF were
higher while HF(NU) was lower in the afternoon ses-
sion as compared to those in the morning session
(p<0.05). In other words, a slight shift towards the
sympathetic activity was found in the afternoon group
before driving. Other authors have reported HRV
differences measured at different hours of the day
[2][3]. This difference does not affect the paired
T-tests shown in Tables 4 and 5 for the effect of driv-
ing, but the authors plan to investigate this effect in the
future studies.
(2). After driving
a. Entire group of physiological parameters
This analysis examined any statistical difference
between the morning and the afternoon session based
upon the entire group of physiological parameters
taken after driving. The results (not shown here for
brevity) indicated that there was no significant group
difference (Wilks’λvalue 0.764; p=0.652) between
the morning and the afternoon session for the 12
physiological parameters taken after driving.
b. Individual physiological parameters (inde-
pendent T-test)
Independent T-tests were utilized to examine in-
dividual physiological parameters for any group dif-
ference between the morning and the afternoon ses-
sion. Results (not shown here) again indicated that no
physiological parameter taken after driving showed a
statistical difference between the morning and the
afternoon session (all p-values>0.05).
(3). Changes in physiological parameters due to
driving
a. Entire group of physiological parameters
changes
This analysis was based upon the entire group of
physiological parameter changes due to driving
measured in the study (a total of 12, see Table 4 or 5)
to examine any statistical difference between the
morning and the afternoon session. The results (not
shown here) indicated that no statistical difference was
observed between the morning and the afternoon ses-
sion (Wilks’λvalue 0.609, p=0.142 ).
b. Individual physiological parameter changes
Examination of individual physiological pa-
rameter changes due to driving for any group differ-
ence between the morning and the afternoon session
was conducted by independent T-tests. Results are
shown in Table 8.
Table 8. Comparison between sessions for individual
physiological parameter changes
Average of change b
due to driving
Parame-
ters Morning Afternoon
Difference
c between
session
Stan-
dard
devia-
tion
DF ap-value
SYS -0.05 7.30 7.35 3.54 38 0.045
DIA -0.15 -0.05 0.10 0.58 38 0.865
HR 4.40 4.40 0.00 1.94 38 1.000
HRV -14.20 -7.60 6.60 3.62 38 0.076
HF (AU) -11.50 -231.50 -220.00 132.07 38 0.104
HF (NU) 8.05 -9.75 -17.80 4.67 38 0.001
LF (AU) -248.25 61.85 310.10 176.87 38 0.088
LF (NU) -8.05 9.75 17.80 4.67 38 0.001
VLF (AU) -1223.80 -653.35 570.45 325.57 38 0.088
LF/HF -0.24 0.68 0.91 0.30 38 0.004
TLP 1.09 0.97 -0.12 0.44 38 0.779
TRP 0.92 0.81 -0.11 0.44 38 0.806
a: DF (degree of freedom)
b: Change was defined as value (after driving) – value
(before driving)
c: Difference was defined as value (afternoon ses-
sion) – value (morning session)
From Table 8, changes in SYS, HF(NU),
LF(NU), and LF/HF were significantly different be-
tween the two driving sessions (p<0.05). Further
analysis shows that all four parameter changes are in
the opposite trend (increase vs. decrease or vise versa)
between the two groups.
3.2.2 Two-way MANOVA (driving session
and driving task) on physiological pa-
rameters
The following analyses treated the driving ses-
sion and the driving task as two independent variables
and investigated each individual effect as well as in-
teractions between variables on physiological pa-
rameters (the dependent variables).
(1). Entire group of physiological parameter
changes
Based upon the entire group of physiological
parameters, MANOVA results are shown in Table 9.
Table 9. MANOVA of driving session and driving
task on entire group of physiological parameters
Effect Test Value F Hypothesis
DF a
Error
DF ap-value
Driving session 0.84 1.13 11 66 0.348
Driving task b0.65 3.15 11 66 0.002
Driving session
× Driving task
Wilks'
λ 0.84 1.12 11 66 0.360
a: DF (degree of freedom)
b: Driving task was defined as value (after driving
task) – value (before driving task)
From Table 9, Driving hour of the day (driving
session) had no significant effect on the entire group of
physiological parameters (p>0.05). In contrast, the
driving task had a significantly effect (p<0.05). As
expected, no significant interaction was observed
between the driving hour and the driving task
(p>0.05).
(2). Individual physiological parameter changes
before and after driving
Independent T-tests were used to assess the effect
of the driving task on each physiological parameter
regardless of the driving session. The results are
shown in Table 10.
Table 10. Comparison of individual physiological
parameters between before- and after-driving
Average (both
sessions included)
Parameters Before
driving After
driving
Difference b
between
before- and
after-driving
Stan-
dard
devia-
tion
DF
ap-value
SYS 116.23 112.60 -3.62 1.77 78 0.044
DIA 73.68 73.78 0.10 0.45 78 0.825
HR 70.80 66.40 -4.40 2.18 78 0.048
HRV 46.63 57.53 10.90 3.69 78 0.004
HF (AU) 545.43 666.93 121.50 109.08 78 0.269
HF (NU) 45.25 46.10 0.85 3.53 78 0.811
LF (AU) 674.10 767.30 93.20 127.28 78 0.466
LF (NU) 54.75 53.90 -0.85 3.53 78 0.811
VLF (AU) 1198.05 2136.63 938.57 274.84 78 0.001
LF/HF 1.62 1.40 -0.22 0.21 78 0.298
TLP 36.62 35.66 -1.02 0.30 78 0.001
TRP 36.69 35.83 -0.86 0.31 78 0.008
a: DF (degree of freedom)
b: Difference was defined as value (after driving) –
value (before driving)
From Table 10, SYS, HR, HRV, VLF(AU), TLP ,
and TRP all showed significant difference between
before- and after-driving (all p-values<0.05). Among
them, HRV and VLF(AU) were higher while SYS,
HR, TLP and TRP were lower after driving.
3.3 Subjective questionnaire analyses
It was confirmed by verbal communication that
all volunteers followed the pre-test instructions (hav-
ing sufficient sleep in the previous night and no eating,
drinking, or exercise one hour prior to the test and
maintained fresh and healthy when reporting to the test
laboratory). In addition, each volunteer filled the fa-
tigue questionnaire before and after driving. Results of
the written questionnaire and statistical analyses are
tabulated in Tables 11 through 14.
3.3.1 Pair T test of questionnaire for morning
and afternoon session
Paired T-tests were conducted for both the
morning and afternoon sessions before and after the
driving task. Results are shown in Tables 11 and 12.
(1). Morning session
Table 11. Paired T test of questionnaire before and
after driving for morning session
Average Score
Question items Before
driving After
driving
Standard
devia-
tion DF ap-value
Body tiredness 2.80 4.75 0.358 19 0.000
Loss of concentration 2.80 4.90 0.315 19 0.000
Desire to lie down 2.70 4.70 0.502 19 0.001
Anxiety 2.15 3.95 0.394 19 0.000
Lack of Energy 2.20 4.15 0.373 19 0.000
Mental response slowdown 2.15 4.05 0.354 19 0.000
Headache 1.15 1.95 0.277 19 0.009
Shoulder stiffening 1.60 3.55 0.425 19 0.000
Waist pain 1.45 2.45 0.251 19 0.001
Lower body numbness 1.30 4.85 0.407 19 0.000
Eye fatigue 2.35 5.35 0.355 19 0.000
Feeling of sleepiness 2.35 5.15 0.367 19 0.000
Feeling of vomit 1.20 1.85 0.334 19 0.067
Hand and foot trembling 1.25 1.95 0.272 19 0.019
Average score per question 1.96 3.82 0.188 19 0.000
Total average score 27.45 53.60 2.638 19 0.000
a: DF (degree of freedom)
From Table 11, all fatigue questions ex-
cept ”Feeling of vomit” showed increased scores after
driving for the morning session (p<0.01 or 0.05). In
addition, average scores and total scores were all sig-
nificantly higher following driving (p<0.01). The
questionnaire results clearly indicated subjective
feeling of driving fatigue expressed by these young
volunteers after 90-min continuous driving.
(2). Afternoon session
Table 12. Paired T test of questionnaire before and
after driving for afternoon session
Average Score
Question items Before
driving After
driving
Standard
devia-
tion DF ap-value
Body tiredness 2.35 4.90 0.366 19 0.000
Loss of concentration 2.45 5.45 0.340 19 0.000
Desire to lie down 2.45 5.40 0.407 19 0.000
Anxiety 2.20 4.55 0.466 19 0.000
Lack of Energy 2.65 4.75 0.306 19 0.000
Mental response slowdown 2.35 4.75 0.302 19 0.000
Headache 1.55 2.45 0.383 19 0.030
Shoulder stiffening 2.40 3.95 0.413 19 0.001
Waist pain 2.30 3.10 0.569 19 0.176
Lower body numbness 1.95 5.05 0.409 19 0.000
Eye fatigue 2.15 5.50 0.350 19 0.000
Feeling of sleepiness 2.35 5.90 0.343 19 0.000
Feeling of vomit 1.35 1.70 0.254 19 0.185
Hand and foot trembling 1.85 2.45 0.319 19 0.076
Average score per question 2.16 4.27 0.228 19 0.000
Total average score 30.35 59.9 3.200 19 0.000
*: DF (degree of freedom)
From Table 12, 11 fatigue questions (out of 13
total questions) showed increased scores after driving
for the afternoon session (p<0.01 or 0.05). The ex-
ceptions were “Waist pain”, “Feeling of vomit, and
“Hand and foot trembling”. In addition, average scores
and total scores were all significantly higher following
driving (p<0.01). Similar to the morning session,
volunteers in the afternoon session also clearly felt
driving fatigue after 90-min continuous driving.
3.3.2 Independent T test for two sessions be-
fore driving
Table 13. Comparison between sessions for fatigue
scores before driving
Average score
before driving
Question items Morn-
ing
(Group
A)
After-
noon
(Group
B)
Differ-
ence b
b
etween
session
DF
ap-value
Body tiredness 2.80 2.35 -0.45 38 0.247
Loss of concentration 2.80 2.45 -0.35 38 0.351
Desire to lie down 2.70 2.45 -0.25 38 0.561
Anxiety 2.15 2.20 0.05 38 0.884
Lack of Energy 2.20 2.65 0.45 38 0.212
Mental response slowdown 2.15 2.35 0.20 38 0.547
Headache 1.15 1.55 0.40 38 0.159
Shoulder stiffening 1.60 2.40 0.80 38 0.033
Waist pain 1.45 2.30 0.85 38 0.021
Lower body numbness 1.30 1.95 0.65 38 0.039
Eye fatigue 2.35 2.15 -0.20 38 0.615
Feeling of sleepiness 2.35 2.35 0.00 38 1.000
Feeling of vomit 1.20 1.35 0.15 38 0.466
Hand and foot trembling 1.25 1.85 0.60 38 0.023
Average score per question 1.96 2.16 0.20 38 0.393
Total average score 27.45 30.35 2.90 38 0.393
a: DF (degree of freedom)
b: Difference was defined as value (afternoon ses-
sion) – value (morning session)
From Table 13, average score per question and
total average score were similar between the two ses-
sions measured before driving, as examined by inde-
pendent T-tests. None of the 13 questions showed a
p-value below 0.01. Overall, volunteers were not
feeling fatigue (all scores below 3.0, the threshold for
feeling of fatigue, see definition in Table 3) before
driving regardless of driving hour.
3.3.3 Independent T test for driving-induced
changes in fatigue scores between ses-
sion
Table 14. Comparison between sessions for fatigue
score changes due to driving
Average of
change b due to
driving
Question items Morn-
ing
(Group
A)
After-
noon
(Group
B)
Differ-
ence c
between
session
DF
ap-value
Body tiredness 1.95 2.55 0.60 38 0.249
Loss of concentration 2.10 3.00 0.90 38 0.060
Desire to lie down 2.00 2.95 0.95 38 0.150
Anxiety 1.80 2.35 0.55 38 0.374
Lack of Energy 1.95 2.10 0.15 38 0.758
Mental response slowdown 1.90 2.40 0.50 38 0.290
Headache 0.80 0.90 0.10 38 0.834
Shoulder stiffening 1.95 1.55 -0.40 38 0.504
Waist pain 1.00 0.80 -0.20 38 0.750
Lower body numbness 3.55 3.10 -0.45 38 0.441
Eye fatigue 3.00 3.35 0.35 38 0.487
Feeling of sleepiness 2.80 3.55 0.75 38 0.144
Feeling of vomit 0.65 0.35 -0.30 38 0.480
Hand and foot trembling 0.70 0.60 -0.10 38 0.813
Average score per question 1.86 2.11 0.24 38 0.417
Total average score 26.15 29.55 3.40 38 0.417
a: DF (degree of freedom)
b: Change was defined as value (after driving) – value
(before driving)
c: Difference was defined as value (afternoon ses-
sion) – value (morning session)
Examination for fatigue score changes due to
driving by independent T-test showed that there was
no statistic difference between the morning and af-
ternoon session (Table 14). None of the 13 fatigue
questions showed significant difference in terms of
score change between the two driving sessions. Av-
erage score changes per question as well as total score
changes also showed no significant difference due to
driving hour. It is concluded that increase in fatigue
score due to driving was not significantly affected by
driving hour (morning or afternoon). Although the
overall session difference did not reach statistical
significance, the data details seem to suggest that the
afternoon volunteers felt slightly more tired due to
driving. Among the 13 questions, the afternoon vol-
unteers felt more fatigued (based upon the score
change) for 9 questions and less fatigued for 4 ques-
tions than the morning volunteers. Both average score
change and total average score change were higher for
the afternoon session. As known by scientists, sub-
jective questionnaire is often not as accurate or reliable
as other quantitative means of measurement. More
data or refined fatigue questions are needed in this area
in the future research.
4. DISCUSSION
Multiple vital physiological parameters were
monitored in the study to evaluate the effect of driving
task. In general, for driving tasks conducted in the
morning (such as Group A in the study), drivers are in
a fresh state both physically and mentally, ANS re-
sponse to the driving task should be swift and satis-
factory. On the other hand, the sleepy and tiring body
may have difficulty in coping with driving fatigue for
driving in the afternoon (such as Group B in the
study). Our results tend to agree with the expected
effect. Each physiological paramer change is
discussed below.
4.1 Blood pressures (SYS and DIA)
From Tables 4 and 5, SYS was almost unchanged
for Group A in the morning session after driving. In
contrast, a reduction in SYS was observed for Group B
tested in the afternoon. Diastolic pressure was little
changed for either session. Static driving task ac-
companied by lower body (waist, hip, abdomen, and
legs) immobilization resulted in poor systemic circu-
lation. ANS reacted to the physiological change
swiftly but differently in each session. For Group A,
homeostasis of blood pressures was maintained
through the activation of sympathetic nervous system.
For Group B, activity of sympathetic nervous system
slowed down while parasympathetic branch activity
increased (see discussions on HRV related parameters
below). As a result, the body entered a sleepy state and
SYS was lowered. Garde et al. [5], Hjorskov et al. [7],
and Veltman et al. [25] all observed the active role
played by the sympathetic branch in the modulation of
blood pressure during work load. An increase in LF
was often accompanied by blood pressure rise. From
Tables 4 and 5, activation of the sympathetic branch
was a body response to maintain blood pressures in the
morning session while de-activation of the sympa-
thetic branch along with activation of the parasym-
pathetic branch in the afternoon session prepared the
body to enter a restful (or sleepy) state (see more
discussions on HRV parameters later).
In contrast, Fumio et al. [4] conducted blood
pressure and HRV tests on city taxi drivers and ob-
served increased blood pressure during service. There
are two major differences in the experimental setting
between their and our study. For taxi drivers, their
physical and mental stresses are much higher than
those experienced in the indoor simulation. Even
though our driving room was designed to duplicate
road driving as closely as possible, volunteers might
regard the test as a non-real driving event. In our
study, volunteers were not allowed to get up or leave
the driving seat in the 90-min continuous driving. City
taxi drivers, on the other hand, need to get out to open
doors or handle luggage from time to time. The extent
of lower body immobilization is thus lower for taxi
drivers. Driving conditions in our simulation are more
similar to those encountered in long-distance highway
driving. In cardiology and hemodynamics, pulse
pressure is defined as the difference between systolic
pressure and diastolic pressure. In this study, since
diastolic pressure was little changed after driving
(Tables 4 and 5), it follows that any change in systolic
pressure is directly related to pulse pressure change. In
general, a lower pulse pressure represents a lower
cardiac output. Furthermore, decrease in cardiac out-
put is seen when the body demand for energy is low or
the body is “exhausted”. Thus, a reduction in pulse
pressure (as seen in the afternoon session in this study)
might be one of the candidates in terms of physio-
logical parameter that can be employed to quantify
driving fatigue.
4.2 Heart rate (HR)
From Tables 4 and 5, HR reduced significantly in
both Group A and B after driving. During driving,
body movement was limited with the hip, legs, and
feet almost stationary. The inactivity in the lower body
(limbs and abdomen) also caused poor circulation. In
particular, venous return to the heart was slowed
down. As a result, HR was lowered from the restful
condition. Several previous studies Lal et al. [10], Li et
al. [14][17], and Miloevic [21] on long-hour driving
also reported a reduction in HR, consistent with our
finding. Again, based upon cardiology and hemody-
namics, a slower heart rate represents a smaller cardiac
output. Decrease in cardiac output may be a sign that
the body is “tired”. Thus, a reduction in HR might be
one of the candidates in terms of physiological pa-
rameter that can be employed to quantify driving fa-
tigue.
4.3 Heart rate variability (HRV)
From Tables 4 and 5, HRV increased signifi-
cantly in both morning and afternoon sessions after
driving. Li at al. [14] and Yang et al. [27] also ob-
served similar trends in their driving studies. Further
analysis indicates that the increase in HRV for the
morning session (from 44.5ms to 58.7ms) was con-
tributed mainly from the sympathetic activity LF and
very low frequency VLF (whose physiological
meaning not clear). In contrast, the afternoon session
saw HRV increase (from 48.7ms to 56.3ms) coming
mainly from the parasympathetic activity HF and very
low frequency VLF. Literature reports from Garded et
al. [5], Hjorskov et al. [7], Jorna [9] and Veltman et al.
[25] often observed a decrease in HRV after subjects
conducted operations involving complexity or ur-
gency. On the other hand, HRV was seen increased
after a light, routine, or near-restful operation. It is
noted that no constraint in body movement was en-
countered in those studies. Another factor affecting the
ANS response is that indoor simulated driving is not as
intense as the actual road driving. Also noted is that
HRV represents the total activity of ANS in a general
sense. Body response to a specific action may take
different routes (branches) to achieve the new balance.
For instance, to lower the heart rate or blood pressure,
one route would be to lower the sympathetic branch
activity. Another route is to increase the activity of
parasympathetic branch. The first route will decrease
HRV while the second will do the opposite. In certain
occasions or subjects, both routes may be taken to-
gether to achieve the adjustment. Thus, it is important
to examine not only the change in HRV but also pa-
rameters related to each branch of ANS, as discussed
below. Due to the complexity involved, it is a general
belief by the authors that HRV alone is not a good
candidate in terms of physiological parameter for
quantifying driving fatigue.
4.4 Sympathetic nerve activity [LF(AU)
and LF(NU)
From Tables 4 and 5, LF (AU) and LF (NU) both
increased significantly after driving for Group A
conducted in the morning. In contrast, LF (AU) de-
creased and LF (NU) decreased significantly for
Group B in the afternoon session. The results show
that the sympathetic nervous activity was enhanced in
Group A but reduced in Group B. Our morning session
results are consistent with previous studies Jiao et al.
[8], Li et al. [14,15,16] and Yang et al. [27] where the
driving tasks were also conducted in the morning and
increases in LF(AU) and LF(NU) were found. Ac-
cording to our results, subjects in the morning were in
a fresher body state so that activation and execution of
sympathetic nervous system was more satisfactory
compared to the afternoon driving. As a result, ho-
meostasis of systolic pressure was maintained in the
morning session. For the afternoon session, driving
fatigue (discussed later in Questionnaire results) must
have reached a critical level that subjects “turned off”
the sympathetic nervous system and lowered the sys-
tolic pressure to some extent in order to force the body
to rest.
4.5 Parasympathetic nerve activity
[HF(AU) and HF(NU)]
From Tables 4 and 5, HF (AU) held almost con-
stant and HF (NU) decreased [due to increased LF
(AU)] after driving for the morning session. In con-
trast, Group B in the afternoon session exhibited in-
creased HF (AU) (non-significant) and HF (NU) (sig-
nificant). The results show that in a relative sense (in
terms of normalized units) the parasympathetic nerv-
ous activity was reduced in the morning (Group A) but
enhanced in the afternoon (Group B) after driving.
Again, our morning finding on HF(NU) decrease is in
agreement with previous driving studies reported by
Jiao et al. [8], Li et al. [14,15,16] and Yang et al. [27]
conducted in the morning. It is noted that for the
morning driving when the body was in a fresh state,
the parasympathetic nervous branch was in sync (de-
crease in activity) with the sympathetic nervous
branch (increase in activity) to collectively boost car-
diac output and maintain homeostasis during driving.
In contrast, the exhausting body during the afternoon
driving reduced the sympathetic activity (discussed
earlier) and enhanced the parasympathetic activity
(again in sync) to enter a sleepy mode.
4.6 Very low frequency [VLF(AU)]
From Tables 4 and 5, VLF (AU) increased sig-
nificantly for both the Group A in the morning session
and Group B in the afternoon. The physiological
meaning of VLF (AU) is not defined in the 1996
Standard [23].
4.7 Sympatho-vagal balance index
(LF/HF)
From Tables 4 and 5, a significant increase in
LF/HF was observed in the morning session for Group
A attributing to higher sympathetic activity (LF) and
almost unchanged parasympathetic activity (HF). In
contrast, LF/HF exhibited a decrease in the afternoon
session for Group B as a result of unchanged sympa-
thetic activity (LF) and enhanced parasympathetic
activity (HF). The results show that after driving, the
balance of ANS activities shifted towards the sympa-
thetic branch for the morning session but parasympa-
thetic branch for the afternoon session. The objective
of morning ANS activities in the body was to increase
cardiac output and maintain homeostasis of vital signs
(such as systolic pressure) while ANS activities in the
afternoon called for the body to rest. Again, our
morning results on increased LF/HF are consistent
with previous driving studies published by Jiao et al.
[8], Li et al. [14,15,16] and Yang et al. [27] conducted
in the morning. It is well known that the parasympa-
thetic tone is dominant when the body is under a
restful or sleepy state. A lower LF/HF (or a shift to-
wards the parasympathetic branch from baseline)
might be a sign that the body is fatigued. Thus, a re-
duction in LF/HF might be one of the candidates in
terms of physiological parameter that can be employed
to quantify driving fatigue.
4.8 Temperature of palm
From Tables 4 and 5, both sessions showed sig-
nificant drops in palm temperatures after driving. The
results clearly show that body temperature, especially
one measured at extremity, is a sensitive indicator for
blood circulation and homeostasis. The results also
show that even with successful activation of sympa-
thetic branch in the morning, poor blood circulation
due to lower body immobilization was not fully com-
pensated. Again, a lower body temperature might be
an early sign that the body can not cope with poor
circulation and it is about to be or already fatigued.
Thus, a reduction in body temperature might be one of
the candidates in terms of physiological parameter that
can be employed to quantify driving fatigue.
4.9 Subjective questionnaire
From Tables 11 through 14, questionnaire results
on driving fatigue clearly showed that 90-min con-
tinuous driving resulted in multiple symptoms of body
and mental fatigue, regardless of driving hour (morn-
ing or afternoon). Statistically, there was no difference
in terms of fatigue score changes due to driving be-
tween the morning and afternoon session. However,
from measured physiological parameters, distinct
difference was seen between the two driving session in
terms of changes in systolic pressure and HRV pa-
rameters (such as LF, HF or LF/HF). While more
studies are needed to provide a firm interpretation, it is
a general belief that the subjective questionnaire is not
as a reliable tool as the direct measurement of
physiological parameters for assessment of body con-
ditions.
4.10 Overall discussion
When defining fatigue, it is important to discuss
the association between body fatigue and boredom
(monotony), since one may affect or instigate the
other. In our preliminary driving studies, no differen-
tiation was made between body fatigue and boredom.
We considered boredom as a special type of fatigue
due to the fact that boredom is caused by a reduction of
the activation level of the brain [10]. Also, our indoor
driving simulation is close to driving conditions on a
highway with monotonic scenes and routine vehicle
maneuver. Long-hour driving under these conditions
will involve both body fatigue as well as boredom, as
our physiological data and questionnaire results have
shown.
As driving fatigue develops, cardiovascular sys-
tem is unable to fulfill the basic physiological needs.
Hands and feet become cold, heart rate slows
[10][14][17][21], and eventually blood pressures go
down. Poor circulation causes muscle pain and
numbness. Hypoxia in brain induces drowsiness and
loss of concentration. Under these circumstances, our
body re-assesses the new physiological needs and
activates ANS instantly (seen with HRV increase in
the study) [14][27]. If re-assessment finds the body in
the state of exhaustion which requires an immediate
rest, parasympathetic branch in ANS is called upon
(seen with Group B in this study). On the other hand, if
re-assessment determines that a boost to cardiovas-
cular system can meet the new physiological needs,
sympathetic branch is activated (Group A in this
study) [8][14][15][16][27]. The success of ANS action
depends on several factors, including mental alertness
[5][7][11][25], heart muscle strength, and peripheral
circulation resistance. Immobilization of the lower
body (limbs, abdomen, and hip), which increases
blood flow resistance significantly, could compromise
the sympathetic nerve’s function in boosting systemic
circulation. Since the monitoring of physiological
parameters was done at the beginning and the ending
of the driving task in the study, it is unclear whether or
not the sympathetic branch was activated first in the
early face of driving for the afternoon session (Group
B). Note that at the end of driving for this group, the
parasympathetic activity was enhanced. Under these
conditions, if the driver continues to drive, the risk of
traffic accidents would be extremely high. For Group
A conducted in the morning, the sympathetic nerve
was enhanced at the end of driving
[8][14][15][16][27], which helped to maintain systolic
pressure and other vital parameters. But even the ANS
action in this group is regarded as a partial success
since palm temperatures and heart rate were still below
baseline [14][17].
The results of this preliminary study indicate that
driving fatigue can be tracked by vital physiological
parameters, including extremity temperatures, heart
rate, and blood pressures. Significant deviation of any
vital parameter from baseline during or after driving
should be viewed as a symptom of fatigue. In dis-
agreement with some previous authors, we regard
HRV increase as a response of ANS to driving (or any
body) activity, and not necessarily related to driving
fatigue. We further propose that among the HRV
parameters measured, parasympathetic nerve indexes
[HF (AU) and HF (NU), expected to have a positive
correlation with fatigue] and sympatho-vagal balance
index LF/HF (expected to have a negative correlation
with fatigue) are two most promising candidates that
could be employed for quantitative ranking of driving
fatigue.
5. CONCLUSION
Driving is a demanding task both mentally and
physically. Driving fatigue remains one of the most
common causes for traffic accidents. Immobilization
of the lower body is thought to play a major role in
driving fatigue, as it hinders blood circulation and
induce hemodynamic changes. ANS activation may
overcome some fatigue symptoms, but recovery is
often incomplete due to immobilization. Distinct
trends were found between two driving sessions in the
study. In the morning session, driving caused decrease
in palm temperatures and heart rate, but blood pres-
sures were maintained by activation of the sympathetic
nervous system. For the afternoon session, in sharp
contrast, palm temperatures, heart rate, and systolic
pressure were all lowered. Parasympathetic (rather
than sympathetic) nerve was activated prompting the
body to enter a sleepy state, which greatly increases
driving accident risks. Monitoring of physiological
parameters in the study had gained great insight into
mechanisms of homeostasis and provided a foundation
in the future work to quantify driving fatigue based
upon deviation from homeostatic states. For the first
time in literature, in the authors’ belief, ANS effects
was shown clearly to be related to driving fatigue,
homeostasis, and traffic accident risks.
6. FUTURE RESEARCH
All the volunteers involved in the study were
young college students with active ANS and a healthy
cardiovascular system. Yet, significant deviation from
homeostasis was observed after 90-min simulated
driving. In the future work, we will recruit older and
less healthy volunteers to assess body response under
worsen driving fatigue conditions. We will conduct
driving tests during the evening hours also. Further-
more, remedy techniques for the reduction of driving
fatigue (such as exercise break, body massage, …etc.)
will be evaluated. A score-based driving fatigue index
is to be developed in the future work.
ACKNOWLEDGMENT
Financial support by National Science Council
(Project No. 95-2221-E-007-176) is acknowledged.
Special thanks are given to all co-workers who par- ticipated in the project.
REFERENCES
1. Carrasco, S., R. Gonzalez, J. Jimenez, R. Roman, V.
Medina and J. Azpiroz, “Comparison of the heart rate
variability parameters obtained from the electrocar-
diogram and the blood pressure wave”, Journal of
Medical Engineering & Technology, 22, 195-205
(1998).
2. Fauchier, L., D. Babuty, M. L. Autret, P. Poret, P.
Cosnay and J. P. Fauchier, “Influence of duration and
hour of recording on spectral measurement of heart
rate variability”, Journal of the Autonomic Nervous
System, 73, 1-6 (1998).
3. Freitas, J., P. Lago, J. Puig, M. J. Carvalho, O. Costa
and A. F. de Freitas, “Circadian heart rate variability
rhythm in shift workers”, Journal of Electrocardiol-
ogy, 30(1), 39-44 (1997).
4. Fumio, K., W. Takemasa, W. Misuzu, A. Yasuhiro, T.
Teruyuki, N. Taisuke, F. Hikari, T. Kiyomi, O.
Akiyoshi, O. Rumi and H. Junichiro, “Blood pressure
and heart rate variability in taxi drivers on long duty
schedules”, Journal of Occupational Health, 44,
214-220 (2002).
5. Garde, A. H., B. Laursen, A. H. Jorgensen and B. R.
Jensen, “Effects of mental and physical demands on
heart rate variability during computer work”, Euro-
pean Journal of Applied Physiology, 87, 456-461
(2002).
6. Giardino, N. D., P. M. Lehrer and R. Edelberg,
“Comparison of finger plethysmograph to ECG in the
measurement of heart rate variability”, Psychophysi-
ology, 39, 246-253 (2002).
7. Hjortskov, N., D. Rissen, A. K. Blansted, N. Fallen-
tin, U. Lundberg and K. Sogaard, “The effect of
mental stress on heart rate variability and blood
pressure during computer work”, European Journal
of Applied Physiology, 92, 84-89 (2004).
8. Jiao, K., Z. Y. Li, M. Chen, C. Wang and S. Qi, “Ef-
fect of different vibration frequencies on heart rate
variability and driving fatigue in healthy drivers”,
International Archives of Occupational and Envi-
ronment Health, 77, 205-212 (2004).
9. Jorna, P. G. A. M., “Heart rate and workload varia-
tions in actual and simulated flight”, Ergonomics,
36(9), 1043-1054(1993).
10. Lal, S. K. L. and A. Craig, “A critical of the psy-
chophysiology of driver fatigue”, Biological Psy-
chology, 55, 173-194 (2001).
11. Lee, Y. H. and B. S. Liu, “In-flight workload as-
sessment: comparison of subjective and physiological
measurements” Aviation, Space, and Environmental
Medicine, 74(10), 1078-1084 (2003).
12. Liang, W. C., J. Yuan, D.C. Sun and M. H Lin,
“Variation in physiological parameters before and
after an indoor simulated driving task in the morn-
ing”, Journal of Industrial Technology and Man-
agement, 1, 261-281 (2006).
13. Liao, D. P., R. W. Barnes, L. E. Chambless, R. J.
Simpson, P. Sorlie and G. Heiss, “Age, race, and sex
difference in autonomic cardiac function measured by
spectral analysis of heart rate variability - the ARIC
study”, The American Journal of The Cardiology, 76,
906-912 (1995).
14. Li, Z.Y., C. T. Wang, F. T. Arthur Mak and H. K.
Daniel Chow, “Effect of acupuncture on heart rate
variability in normal subjects under fatigue and
non-fatigue”, European Journal of Applied Physiol-
ogy, 94, 633-640 (2005).
15. Li, Z.Y., K. Jiao, M. Chen and C. T. Wang, “Effect of
magnitopuncture on sympathetic and parasympa-
thetic nerve activities in healthy drivers – assessment
by power spectrum analysis of heart rate variability”,
European Journal of Applied Physiology, 88,
404-410 (2003a).
16. Li, Z.Y., K. Jiao, M. Chen and C. T. Wang, “Power
spectrum analysis of heart rate variability of driver
during simulated driving”, Chinese Journal of Bio-
medical Engineering, 22(6) (2003b).
17. Li, Z.Y., K. Jiao, M. Chen and C. T. Wang, “Reduc-
ing the effects of driving fatigue with magnitopunc-
ture stimulation”, Accident Analysis and Prevention,
36, 501-505 (2004).
18. Mark S. Sanders and Ernest J. McCormick, Human
Factors in Engineering and Design (7th Edition),
McGraw-Hill Inc. Taiwan Taichung, 818-819 (1992).
19. Martini and Bartholomew, Essentials of Anatomy and
Physiology, Frederic H. Martini, Inc., Taiwan Taipei,
6-14 (2003).
20. McKinley, P. S., P. A. Shapiro, E. Bagiella, M. M.
Myers, R. E. De Meersman, I. Grant and R. P. Sloan1,
“Deriving heart period variability from blood pres-
sure waveforms”, Journal of Applied Physiology, 95,
1431-1438 (2003).
21. Milo evic, S., “Drivers’ fatigue studies”, Ergo-
nomics, 40(3), 381-389(1997).
22. Sun, D. C., “ANSWatch wrist monitor technical
report and ANSWatch wrist monitor pre-clinical test
report”, both submitted to Taiwan DOH (2005).
23. Sun, D.C., Guo Y.Y., Liang Y.H., Tang S.H, Lin
Y.F., Chang Y.X., Lin Z.G., Introduction of a new
wrist patient monitor - ANSWatch and its clinical
applications, Proceeding for the 9th Conference on
Engineering Technology and Application of Chinese
and Western Medicine, May. 33-38, Taichung, Tai-
wan (2006).
24. Task Force of the European Society of Cardiology
and the North American Society of Pacing and Elec-
trophysiology, “Heart rate variability：standards of
measurement, physiological interpretation, and
clinical use”, Circulation, 93, 1043-1065 (1996).
25. Veltman, J. A. and A. W. K. Gaillard, “Physiological
workload reactions to increasing levels of task diffi-
culty”, Ergonomics, 41(5), 656-669(1998).
26. Wahlstrom, J., P. W. Hagberg, J. Johson, D. Svensson
and Rempel, “Influence of time pressure and verbal
provocation on physiological and psychological re-
actions during work with a computer mouse”, Euro-
pean Journal of Applied Physiology, 87, 257-263
(2002).
27. Yang, Y. S., Z. Q. Yao, Z. Y. Li, K. Jiao, “Investiga-
tion on correlation between ECG indexes and driving
fatigue”, Machinery Design Manufacture, 5, 94-95
(2002).
28. Zhang, J., “Effect of age and sex on heart rate vari-
ability in healthy subjects”, Journal of Manipulative
and Physiological Therapeutics, 30, 374-379 (2007).
ABOUT THE AUTHORS
Wen-Chieh Liang is currently a PhD student at De-
partment of Industrial Engineering & Engineering
Management, National Tsing-Hua University, Tai-
wan; Visiting Lecturer at Department of Industrial
Education and Technology, National Chang-Hua
University of Education, Taiwan; Department of In-
dustrial Engineering and Management, Nan-Kai In-
stitute of Technology, Taiwan; and Department of
Industrial Engineering and Management, National
Chin-Yi University of Technology, Taiwan.
John Yuan is a professor in the Department of In-
dustrial Engineering & Engineering Management,
National Tsing-Hua University, Taiwan. Dr. Yuan
received his PhD degree from Institute of the Mathe-
matics, University of Tulane, New Orleans, USA.
D. C. Sun is currently a General Manager of Taiwan
Scientific Corporation, Taipei, Taiwan. A former
R&D Director at Pfizer Inc., New York, and a tech-
nical consultant at Industrial Technology Research
Institute (ITRI), Taiwan, Dr. Sun received his PhD
degree from Department of Chemical Engineering,
University of Pittsburgh, PA, USA.
Ming-Han Lin is a professor in the Department of
Automatic Engineering, Ta-Hwa Institute of Tech-
nology, Taiwan. Dr. Lin received his PhD degree from
Institute of Power Mechanical Engineering, National
Tsing-Hua University, Taiwan.
(Received April 2007; Revised May 2007; Accepted
April 2008)
模擬駕駛任務所引發的生理參數變化：早上 vs.下午時段
梁文杰* 阮約翰
國立清華大學工業工程與工程管理學系
300 新竹市光復路二段 101 號
孫德銓
台灣科學地股份有限公司總經理
林明漢
大華技術學院自動化工程學系
摘要
簡介：駕駛疲勞是造成交通意外事故的一項很常見的原因，下半身包括腿部、臀部及
腰部無法移動於駕駛疲勞中扮演一個很重要的角色，此種情況會阻礙血液循環及引發
血液動力學的的改變。目的：本研究之目的為探討上午及下午兩時段分別執行室內靜
態駕駛模擬，並且監視生理參數的變化。方法：40位年輕男性受測者被分成A(上午)、
B(下午)兩組來進行90分鐘的室內模擬駕駛任務，駕駛任務前後使用新型腕式生理監
視器-心律大師(ANSWatch)來量測血壓(SYS/DIA)、心搏率(HR)、心率變異(HRV)等
生理參數，心律大師(ANSWatch)是直接採用對稱於手腕處內建生物感應器
(bio-sensor)而獲取橈動脈波形，另外藉由高精確度紅外線溫度量測儀來量測手掌溫
度，每一位受測者於駕駛任務前後被要求填寫主觀疲勞問卷評價。結果：(1).由成對T
檢定知：無論上午或下午時段，駕駛任務後皆會造成心搏率(HR)及手掌溫度顯著遞
減、HRV及VLF(AU) 顯著遞增；對上午時段而言，駕駛後LF(AU)及LF(NU) 顯著遞增、
HF(NU) 顯著遞減；反之，對下午時段而言，駕駛後LF(NU)及LF/HF顯著遞減、HF(NU)
顯著遞增(p<0.05)。(2).由One-way及Two-way多變量分析(MANOVA)知：整體而言，
駕駛任務前或後，上下午兩時段受測者生理參數沒有顯著差異，雖然駕駛任務前
LF(AU)、LF(NU)及LF/HF三項參數下午時段高於上午時段(p<0.05)。(3).由主觀疲勞問
卷知：所有受測者於駕駛任務後都明顯感覺疲勞，而駕駛任務前(Score baseline)及駕
駛前後變化(Score change)對兩時段都未達顯著的差異。結論：不同時段駕駛任務後多
項生理參數之變化趨勢有所差異，從上午時段結果知，由於連續駕駛使得下半身血液
循環不良，造成手掌溫度及心搏率(HR)遞減，但由於體內自律神經啟動交感，而使得
LF(AU)、LF(NU)及HRV明顯遞增，因此收縮壓(SYS)得以維持恆定狀態；但到了下午
時段，手掌溫度、心搏率(HR)及收縮壓(SYS)皆明顯遞減，此刻體內啟動副交感
(HF(NU)明顯遞增)而促使身體進入昏睡狀態，此狀況在實際的道路駕駛中會增加意外
的風險，本研究藉由多重生理參數的監控可了解身體維持恆定的程度，並在未來將偏
離恆定狀態的程度定量為駕駛疲勞或昏睡指標之一。
關鍵詞：駕駛疲勞，心率變異，恆定，自律神經系統，血壓，模擬駕駛，ANSWatch
(*聯絡人: richard168@mail.e88.com.tw)
