PERCLOS-based technologies for detecting drowsiness: Current evidence and future directions

Abstract

Drowsiness associated with sleep loss and circadian misalignment is a risk factor for accidents and human error. The percentage of time that the eyes are more than 80% closed (PERCLOS) is one of the most validated indices used for the passive detection of drowsiness, which is increased with sleep deprivation, after partial sleep restriction, at nighttime, and by other drowsiness manipulations during vigilance tests, simulated driving, and on-road driving. However, some cases have been reported wherein PERCLOS was not affected by drowsiness manipulations, such as in moderate drowsiness conditions, in older adults, and during aviation-related tasks. Additionally, although PERCLOS is one of the most sensitive indices for detecting drowsiness-related performance impairments during the psychomotor vigilance test or behavioral maintenance of wakefulness test, no single index is currently available as an optimal marker for detecting drowsiness during driving or other real-world situations. Based on the current published evidence, this narrative review suggests that future studies should focus on: (1) standardization to minimize differences in the definition of PERCLOS between studies; (2) extensive validation using a single device that utilizes PERCLOS-based technology; (3) development and validation of technologies that integrate PERCLOS with other behavioral and/or physiological indices, because PERCLOS alone may not be sufficiently sensitive for detecting drowsiness caused by factors other than falling asleep, such as inattention or distraction; and (4) further validation studies and field trials targeting sleep disorders and trials in real-world environments. Through such studies, PERCLOS-based technology may contribute to preventing drowsiness-related accidents and human error.

Full text

Accepted Manuscript
© The Author(s) 2023. Published by Oxford University Press on behalf of Sleep Research Society.
This is an Open Access article distributed under the terms of the Creative Commons Attribution
License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse,
distribution, and reproduction in any medium, provided the original work is properly cited.
PERCLOS-based technologies for detecting drowsiness: Current evidence and future directions
Takashi Abe1,＊
1International Institute for Integrative Sleep Medicine (WPI-IIIS), University of Tsukuba, Japan
*Corresponding author:
Takashi Abe, Ph.D.
International Institute for Integrative Sleep Medicine (WPI-IIIS), University of Tsukuba, 1-1-1
Tennodai, Tsukuba, Ibaraki 305-8575, Japan.
E-mail: abe.takashi.gp@u.tsukuba.ac.jp
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
Abstract
Drowsiness associated with sleep loss and circadian misalignment is a risk factor for
accidents and human error. The percentage of time that the eyes are more than 80% closed
(PERCLOS) is one of the most validated indices used for the passive detection of drowsiness, which is
increased with sleep deprivation, after partial sleep restriction, at nighttime, and by other
drowsiness manipulations during vigilance tests, simulated driving, and on-road driving. However,
some cases have been reported wherein PERCLOS was not affected by drowsiness manipulations,
such as in moderate drowsiness conditions, in older adults, and during aviation-related tasks.
Additionally, although PERCLOS is one of the most sensitive indices for detecting drowsiness-related
performance impairments during the psychomotor vigilance test or behavioral maintenance of
wakefulness test, no single index is currently available as an optimal marker for detecting drowsiness
during driving or other real-world situations. Based on the current published evidence, this narrative
review suggests that future studies should focus on: (1) standardization to minimize differences in
the definition of PERCLOS between studies; (2) extensive validation using a single device that utilizes
PERCLOS-based technology; (3) development and validation of technologies that integrate PERCLOS
with other behavioral and/or physiological indices, because PERCLOS alone may not be sufficiently
sensitive for detecting drowsiness caused by factors other than falling asleep, such as inattention or
distraction; and (4) further validation studies and field trials targeting sleep disorders and trials in
real-world environments. Through such studies, PERCLOS-based technology may contribute to
preventing drowsiness-related accidents and human error.
Keywords: PERCLOS; psychomotor vigilance test; vigilant attention; slow eyelid closure; sleepiness;
alertness; drowsiness; sleepiness; monitoring; drowsy driving
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
Statement of Significance
Extensive evidence indicates that the percentage of time that the eyes are 80% closed
(PERCLOS) increases with drowsiness caused by sleep loss or circadian misalignment during vigilance
tests, simulated driving, and on-road driving. However, this is not always the case, such as under
moderate drowsiness conditions, in older adults, and during aviation-related tasks. Although
PERCLOS is one of the most accurate measures for detecting drowsiness in vigilance tests, it may not
always be the most reliable indicator for detecting drowsiness while driving or in other real-world
scenarios. Integrating PERCLOS with other physiological or behavioral measures may enable the
robust detection of drowsiness in a variety of real-world situations and contribute to preventing
drowsiness-related accidents and human error.
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
Introduction
Multiple factors are involved in “sleepiness,” including sleep loss/sleep restriction, 1–7 time
since awakening, 6–9 biological rhythms such as circadian, circasemidian, and ultradian rhythms, 8,10–12
motivation, 13–15 sleep inertia, 16–18 body posture, 19 external stimuli such as light, 20 cognitive
workload, 21 and orexin deficiency. 22 Nonlinear interaction between sleep homeostasis and circadian
rhythm also contributes to sleepiness. 6–9,23,24 Sleepiness is a complex phenomenon caused by
multiple factors and their interactions. In addition, elucidating the molecular basis of sleepiness
remains in its early stages. 25,26 Additionally, there is currently no single method that can adequately
measure sleepiness. Although a “physiologic state of sleep need” 27 is a widely accepted definition of
“sleepiness” among sleep researchers and clinicians, current methods typically measure certain
aspects of subjective, behavioral, or physiological changes associated with “sleepiness” on the basis
of an operational definition of sleepiness, such as sleep propensity or drowsiness. 28–34 Sleep
propensity refers to “one’s tendency to fall asleep.” 29–31 Drowsiness is a state that occurs during the
transition between sleep and wakefulness and is associated with changes in cognition, behavior,
vigilance, mood, motivation, autonomic and central nervous system function, other physiological
states, and the subjective experience of sleepiness. 28–32 On the basis of this definition, various
methods for measuring sleepiness have been developed. Because deterioration in cognitive
performance associated with sleepiness is directly related to accidents and human error, 35 the
evaluation of vigilance (defined here as “an ability to sustain attention to a task for a period of time”
36), which is the performance indicator most sensitive to sleepiness, 37,38 is vital for preventing human
error and accidents. The psychomotor vigilance test (PVT) is a representative method for objectively
measuring vigilance relative to sleepiness. 39–42 Although the PVT and its shorter versions 42–44 are
promising methods for measuring vigilance, performing these assays interrupts ongoing tasks by
requiring subjects to respond to stimuli that appear on a device. Therefore, methods for assessing
vigilance without interfering with ongoing tasks are required.
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
Among approaches for passive drowsiness detection, the percentage of time that the eyes
are more than 80% closed (PERCLOS) is one of the most established measures, with a large number
of scientific validation studies and field trials. PERCLOS was first established in 1994 by Wierwille et
al. for detecting driver drowsiness. 45,46 Subsequently, a validation study of PERCLOS for detecting
PVT lapses during sleep deprivation was conducted by Dinges et al. in 1998. 47 The results revealed
that PERCLOS was the most accurate indicator for detecting PVT lapse among several measures,
including electroencephalography (EEG) and blinks. 47 Since then, PERCLOS has been extensively
validated as an index for detecting drowsiness in several situations, including laboratory settings,
simulated driving, and on-road driving. 32,48–50 This narrative review discusses PERCLOS in terms of
validation, comparisons with other indicators, practical applications, limitations, and future
directions for further development.
Methodology for evaluating PERCLOS
PERCLOS can be defined as the percentage of time that the eyes are > 80% closed. 45,47 In
previous studies of PERCLOS, the eye is defined as being closed when the eyelid is < 20% open (0% is
defined as completely closed). 45–47 However, definitions of PERCLOS have vary greatly across
studies. Calculation of PERCLOS determines eye closure based on the degree of instantaneous eyelid
opening. It is possible to set eye closure to 0% or eye-opening to 0%. Previous studies have assumed
complete eye closure at a given instant to be 0%, with (1) 100% representing eyes wide open and 0%
representing eyes closed; 45 (2) the percentage of the distance between eyelids when the diameter
of the iris is represented as 100%; 47,51 and (3) the percentage of the pupil occluded by the eyelid
with the diameter of the pupil was 100%. 52 Eye-opening of less than 25% 53 or 30% 47,54–57 has also
been used as a definition of eye closure in some previous studies. In some cases, to obtain a
measure of PERCLOS in a simplified manner, the eye is considered to be closed (4) when the pupil
cannot be detected, 58,59 or (5) the eye aspect ratio (EAR) 60 is below a definite value. 61 Some studies
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
exclude fast blinks as eye closure, 47,51,52,62 while others include this measure. 58,63 The definitions of
fast eyelid closure include < 250 ms, 47,64 < 400 ms, 52,62 or < 500 ms. 51 The sampling frequencies for
recording the degrees of eyelid opening include 2 Hz, 65 3 Hz, 66 6 Hz, 56 10 Hz, 51,67 24 Hz, 54 60 Hz,
68,69 or 120 Hz. 52,55 A sufficient sampling rate is required when fast blinks are excluded. The analysis
may be performed manually on the basis of the video recording of the eyes 47,51,65 or automatically
by a device. 70–72 In the case of manual analysis, it takes a substantial amount of time to evaluate
each frame. However, automatic measurements may cause discrepancies in values even when
PERCLOS values are simultaneously measured using several instruments. 71 The effects of these
differences in definitions and instruments on the accuracy of vigilance detection require further
study. Initially, the percentage of time with eyes closed (%TEC) measured by infrared reflectance
oculography (i.e., Optalert Drowsiness Measurement System, Sleep Diagnostics Pty Ltd, Melbourne,
Australia) was not referred to as PERCLOS. 73,74 However, a recent study reported %TEC as a measure
of PERCLOS. 75 The %TEC is defined as the “percentage of time that the eyes are deemed closed in
each minute.” 74 This review also discusses %TEC.
What activities does PERCLOS reflect?
PERCLOS can be used to detect decreased vigilance in vigilance tasks using both visual and
auditory stimuli. 76 Thus, the association between PERCLOS and reduced vigilance is not merely
caused by the blockage of the acquisition of external visual information by closing the eyes, but also
by reduced activity of the central nervous system. Neural correlates of slow eyelid closure, a
component of PERCLOS, during sleep deprivation have been reported using functional magnetic
resonance imaging (fMRI), which indicated reduced coupling within the default mode network
(DMN) and weak anticorrelations between the DMN and the dorsal attention network (DAN). 77,78
Furthermore, reduced connectivity within the DMN and anticorrelation between the DMN and DAN
were reported to predict intra-individual temporal fluctuations in response speed during the
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
auditory vigilance task. 78 Similar reductions have been reported under sleep-deprived conditions
during a visual attention task. 79 Thus, such brain activity is likely to be responsible for the
occurrence of slow eyelid closure and the associated reduction in vigilant attention.
Use of PERCLOS under fixation, PVT, and OSLER test
Several studies using tasks that do not necessarily require gaze shifts have repeatedly
demonstrated that PERCLOS shows an increase that is related to drowsiness manipulations (Table 1),
such as sleep deprivation or partial sleep restriction, when conducting the PVT 47,51,52,62 or the Oxford
Sleep Resistance (OSLER) test. 58 The latter is a behavioral test that is similar to the maintenance of
wakefulness test (MWT), which requires a response to a light stimulus occurring every 3 s for 40
min.80 A similar index measured using Optalert, %TEC, has also been shown to increase with
drowsiness manipulation when looking ahead 73 or when conducting PVT. 70,74,81,82 Additionally, the
measurement of PERCLOS during a simple reaction task in the field has captured the disturbance of
circadian rhythms among two types of air traffic controllers: Those who experience fewer night
shifts and those who experience a higher frequency of night shifts. 54 It should be noted that not all
studies have shown an increase in PERCLOS with PVT measurement during sleep deprivation.64,70
The accuracy of PERCLOS in detecting PVT lapse and non-response (response time *RT+ ≥ 3 s)
during the OSLER test associated with drowsiness manipulation is shown to be higher than various
measures, including slow eye movements, blink parameters, pupil diameter, heart rate variability,
and EEG-based measures as demonstrated in Table 2. 47,52,58 While one study reported that %TEC had
stronger correlations with PVT performance than that of PERCLOS measured using Copilot or Johns
Drowsiness Scale (JDS: a composite drowsiness measure of multiple eye metrics using the
Optalert83),70 high accuracy in detecting performance impairment during the PVT and OSLER test is
not always achieved in the %TEC measured with Optalert.73,74,81,84 The cause of this discrepancy
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
between %TEC and PERCLOS is unclear; however, the definition of eye closure or the measurement
device may affect the difference in accuracy between these measures.
Importantly, PERCLOS values during PVT have been shown to be similar in repeated
exposures to sleep deprivation, indicating that the PERCLOS response to sleep deprivation is trait-
like. 62 This is similar to the PVT performance. 85–87 In addition, PERCLOS has been found to exhibit a
higher intraclass correlation coefficient (ICC) compared to various measures, including several
spectral bands of EEG power, heart rate, HRV, blink rate, subjective measures, and PVT measures
(Table 2). 62 The lower intra-individual variation of PERCLOS under the same conditions suggests that
PERCLOS may be more sensitive to changes in sleepiness compared to other indices in the PVT.
A method for detecting vigilance deterioration in PVT with greater accuracy than PERCLOS
has been reported. A PERCLOS-based algorithm that integrates multiple eye metrics was found to
outperform PERCLOS, with ICCs of PVT RT ≥ 300 ms. 51 However, this algorithm requires the use of
electrooculography (EOG) and high-precision eye cameras. Implementing this algorithm with
unobtrusive devices is needed for practical use. In addition, future research should investigate
whether this method is effective in detecting drowsiness-related events in both simulated and real-
world settings, on top of vigilance tests.
Table 1. Summary of the effects of drowsiness manipulation on PERCLOS.
Use of PERCLOS under various driving conditions
Validation of PERCLOS as a measure of drowsiness in operational environments requires
both simulated and field studies.88 Many validation studies for PERCLOS have been conducted,
particularly in the context of driving, with numerous simulation studies reporting increases in
PERCLOS or %TEC during driving after sleep deprivation 57,67,70 or partial sleep deprivation, 59
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
overnight driving, 65,71,89 driving after a night shift, 53 driving at post-lunch dip, 53 driving during sleep
inertia, 90 driving after benzodiazepine administration, 81 or automated driving (combined operation
of adaptive cruise control and lane-keeping assistance) when compared to manual driving. 91 The
effect of automated driving on PERLOS is more pronounced under fatigued conditions (extended
wakefulness or sleep deprivation); 91 PERCLOS has been shown to increase with subjective sleepiness
during simulated driving. 57,90,92 On-road driving studies have also reported increases in PERCLOS
during manual driving with sleep deprivation in younger adults, 72 during nurses’ commutes after a
night shift compared to before a night shift, 93 and during motorway driving at night time.94
There are several issues that need to be addressed in order to increase efficacy of PERCLOS
in detecting drowsiness during on-road driving. First, there are conflicting pieces of evidence
regarding the effectiveness of PERCLOS or %TEC in detecting drowsiness after night shifts. Some
studies have found that PERCLOS or %TEC increases after a night shift when driving on a simulated
course 53 or during an on-road commute. 93 However, other research found no changes in %TEC
during daytime driving on a closed driving track after working a night shift when compared to a night
without working a night shift. 95 Further development and validation of drowsiness monitoring is
urgently needed to reduce higher risk of accidents during driving after a night shift.95–97 Second, it is
necessary to determine the extent to which PERCLOS can detect moderate drowsiness during on-
road driving. PERCLOS in simulated driving was not sensitive to moderate sleep restriction (2–4 h of
sleep reduction). 69 In addition, the relationship between PERCLOS and SD of the lateral lane position
is low when the drivers are in mild and mid-range fatigue conditions.71 However, there are no
studies investigating the effect of moderate sleep restriction on PERCLOS during on-road driving.
Thus, future research should examine the extent to which sleep restriction affects PERCLOS during
on-road driving. Third, there may be limitations to using PERCLOS for detecting drowsiness in older
adults. Studies have shown that PERCLOS in older adults does not always correspond to underload in
automated simulated driving compared to manual driving 91 or deterioration in on-road
performance due to sleep deprivation.72 In the latter study, older drivers showed increased lane
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
deviation as a result of sleep deprivation, suggesting that factors other than falling asleep, such as
inattention and distraction related to drowsiness, 98 may contribute to impairments in driving
performance due to drowsiness. Developing methods to detect drowsiness-related inattention or
distraction may help to improve the low sensitivity of PERCLOS to drowsiness in older adults. Fourth,
there are limited hours in which PERCLOS is more sensitive to drowsiness during on-road driving. A
stimulated driving study showed a higher incidence of PERCLOS in automated driving compared to
manual driving during daytime.91 However, this effect was not observed during on-road driving on
daytime.94 Safety-sensitive real-world driving conditions may reduce the occurrence of drowsiness
due to the underload of the automated driving mode. Further identification of which populations
and situations are more likely to exhibit increased PERCLOS associated with automated driving will
be important in order to improve the utility of PERCLOS in the upcoming need for monitoring
driver’s states during automation mode. Fifth, there are still few studies on the utility of PERCLOS
while driving in patients with sleep disorders. One study measured %TEC during an on-road
naturalistic driving in obstructive sleep apnea group and found no difference in %TEC compared to
healthy controls. 99 More research targeting sleep disorders provides a better understanding of
whether eye metrics, including PERCLOS, are appropriate for assessing fitness-to-drive. 99
What behavioral parameters are related to PERCLOS during drowsy driving? There is
generally robust evidence of a relationship between PERCLOS and measures of lateral position
variability (e.g., number of lane departures or standard deviation (SD) of lateral position) during
simulator driving, 59,65,71,89,100 with the exception of one study. 70 However, this correlation tends to
decrease in low drowsiness conditions. 71 Studies investigating the relationship between PERCLOS
and crashes have produced conflicting results, with some studies finding a correlation 59,65,70 and
others finding no correlation 67,70,81. There are fewer studies that have used braking reaction time as
an indicator, and these studies have also yielded inconsistent results, with some reporting a
correlation65 and others, not 67.
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
Is PERCLOS more sensitive than other indicators in detecting drowsiness during driving? In
vigilance tests, PERCLOS has consistently been shown to be more accurate in detecting drowsiness
than other indicators; however, this is not always the case in the driving situation as shown in Table
2. 59,68,70,71,81,101 Nevertheless, there is still no single index that consistently outperforms the accuracy
of PERCLOS in detecting drowsiness during driving. It will be necessary to gather and consolidate
data from more studies to develop the best measure that accurately detects driving performance
impairments related to drowsiness.
To overcome the issues of PERCLOS described above, integrating PERCLOS with other
physiological or behavioral indicators have been attempted to provide better estimation of driver’s
state or performance during driving, for example: (1) classifying drowsiness categories into three
levels by integrating lane deviation of driving and PERCLOS, 102,103; (2) estimating subjective
sleepiness during simulated driving in three levels (KSS: 1–6; KSS 7; KSS: 8–9) by integrating
PERCLOS, pupil diameter, SD of lateral position, and a steering wheel, with a multilevel ordered logit
model; 92 or (3) models that predict lane-crossing or EEG-defined microsleeps by integrating eye
metrics measured using Optalert (JDS scores, %TEC, and/or AVR), driving performance measures (SD
of lane position, SD of steering wheel position, or steering wheel error), or driver effect. 75 Further
validations of these methods under driving conditions are important in overcoming the current
limitations of PERCLOS.
Table 2. Summary of comparisons of accuracy between PERCLOS and other indicators in drowsiness
detection.
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
Use of PERCLOS under aviation-related conditions
The effectiveness of using PERCLOS for detecting drowsiness in aviation-related tasks is
limited even the task performance showed an effect of extended awake (Table 1). 64 Additionally,
there is an index that showed a higher correlation with the RT during the target acquisition task than
that of PERCLOS (Table 2).64 Consequently, PERCLOS may not be a reliable technique for detecting
performance deterioration caused by drowsiness during aviation-related tasks. 64 However, the same
study did not show an increase in PERCLOS in the PVT associated with sleep deprivation, which has
been shown repeatedly in previous studies. 47,51,52 The authors noted the high variability of PERCLOS
is the cause of low sensitivity. 64 Further research will be required using a more stable technique to
measure drowsiness-related performance impairment during aviation-related tasks.
Use of PERCLOS for sleep detection in patients with sleep disorders
Can PERCLOS be used to detect sleep onset in sleep disorders? A study determining whether
PERCLOS would be a convenient method for detecting sleep during the MWT in patients with sleep
disorders showed that PERCLOS had the best performance in discriminating the occurrence of sleep
in an MWT compared to other metrics (Table 2). 55 However, PERCLOS estimated sleep onset earlier
than that defined by polysomnography (PSG), and the 95% confidence interval was 21.1 min.
Therefore, Kratzel et al. 55 proposed that PERCLOS alone cannot replace the method of detecting
sleep during MWT in clinical settings. However, composite measure of PSG and eyelid closure (a
component of PERCLOS) may improve the usefulness of MWT in clinical settings. In fact, such
combination successfully detects microsleep episodes in the MWT with a time resolution of 1 s.104
Further development of this type of combination may improve the diagnosis, as well as assessment
of treatment and fitness-to-drive by in-depth analysis, of the wake–sleep transition zone.
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
Other applications of PERCLOS
Time-on-task: Several studies have reported the effect of time-on-task on PERCLOS. 51,59,72,94,105,106
The results revealed an time-on-task increase on PERCLOS during PVT, which is more pronounced
from 20 to 34 h after awakening.51 PERCLOS is also affected by time-on-task during driving simulator
tasks 59 or on-road driving. 72,94 The time-on-task effect is particularly pronounced at night in on-road
driving,94 individuals with multiple sclerosis in which daytime sleepiness is a common symptom, 105
or during a task in which the observer is required additional effort to find a target appearing in one
of five locations. 106 PERCLOS is a sensitive index not only for drowsiness, but also for time-on-tasks.
Pre-driving alertness assessments: The utility of PERCLOS as a pre-driving alertness assessment has
been investigated in a study which recorded naturalistic driving over 2 weeks in night shift workers
with pre-driving alertness assessments, including ocular parameters and PVT before driving. The
results revealed that the mean blink duration and PERCLOS were the best discriminators for
predicting the occurrence of behavioral microsleeps, defined as an eye closure duration of > 500 ms,
during subsequent driving. 107 The results indicated the potential for the use of these eye indices for
assessing fitness-to-drive.
Prolonging effect on sleep time: Dinges et al. conducted a vigilance measurement study on
professional truck drivers driving on actual roads. 108 The authors examined the effect of feedback
from driving performance and vigilance measures such as lane tracking performance, PERCLOS, and
PVT. Under conditions with feedback using these metrics, performance during nighttime driving
improved, and non-workday sleep time after returning home increased. Vigilance detection systems
not only inform people of their vigilance status, but may also promote awareness of the importance
of ensuring sleep. 108
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
Hypoxia in cockpits: PERCLOS was recently identified as a potential biomarker for the early stages of
hypoxia in cockpits. 63 PERCLOS responded with the fastest change during which SpO2 was gradually
reduced by 5% in a console located inside a high-altitude hypobaric chamber among the indices
tested. Thus, further studies should be conducted on the utility of PERCLOS as an index for the pilot’s
state in aviation.
Comparing PERCLOS and the Johns Drowsiness Scale (JDS)
The JDS 83,109, like PERCLOS, is one of the most validated drowsiness indices in the field of
sleep research. JDS scores increase with decreased performance during vigilance tasks 73,74,83,84,109
and simulated driving tasks70,83,109–112. Additionally, JDS scores were found to decrease with caffeine
intake, 110,111 decrease with real-time drowsiness feedback,112 and increase with benzodiazepine use.
81 The increased JDS has also been observed during on-road driving after a night shift 93,95 or
extended wakefulness. 113 One of the advantages of the JDS is that, among drowsiness monitoring
devices, validation in a variety of situations using a single device (i.e., Optalert system) is the most
advanced for JDS compared with other measures. 114 In contrast, multiple devices are available for
measuring PERCLOS, and validations with a single device are not as advanced compared with those
for the JDS. 114 It should be noted that there are cases in which a single ocular metric was shown to
be more accurate for detecting decreased PVT performance compared with the JDS. 70,74,84
Whether the JDS or PERCLOS is more accurate as a drowsiness measure is a research topic of
interest. 48,70 A study that simultaneously measured Copilot PERCLOS and Optalert JDS showed no
difference in PERCLOS during PVT or driving tasks after sleep deprivation compared with those after
normal sleep. However, there was a difference in JDS scores and %TEC showing increased values
during both tasks after sleep deprivation compared with those after normal sleep. 70 The lack of
difference in PERCLOS might be due to the limited capability of gaze direction and low sampling (i.e.,
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
3 Hz) of the device. 49 Further research involving direct comparisons between the JDS and PERCLOS
using simultaneous or consecutive measures within a single study are necessary.
Comparing PERCLOS and blink duration
The increase in blink duration has been used to investigate the effects of sleep deprivation,
partial sleep restriction, circadian misalignment, countermeasure of sleepiness and time-on-task on
drowsiness in PVT, 51 simulated driving 115,116, on-road driving 72,94,117,118. A recent review suggested
that blink duration and PERCLOS are the most frequently assessed parameters across studies among
eyelid metrics measured using either an infrared sensor or EOG and consistently increased in
response to extended wakefulness or low circadian alertness, regardless of the task type or
acquisition technique, suggesting that blink duration and PERCLOS are robust measures of
sleepiness. 49
The results of single studies comparing the accuracy of drowsiness detection using PERCLOS
and blink duration are inconsistent among studies. While certain studies suggest that PERCLOS or
%TEC are more accurate than blink duration in detecting drowsiness during the fixation 73, PVT 81,
and OSLER test, 58 other studies indicate that blink duration is more accurate than %TEC in detecting
drowsiness during the auditory PVT 74 and OSLER test 84. Additionally, some studies demonstrate
that blink duration is more associated with crashes during simulated driving 59,81, while others show
that PERCLOS is more accurate in detecting line crossing 59. Sleep detection during MWT is more
accurate using PERCLOS than blink duration 55. Detection of subjective sleepiness is more accurate
using %TEC in the fixation 73 and blink duration in the simulated driving 59. More systematic
comparisons of the accuracies between these indicators in various situations are necessary.
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
Future directions
As noted above, many studies have shown that drowsiness manipulation causes an increase
in PERCLOS during several tasks (Table 1). However, cases in which PERCLOS is not affected by
drowsiness manipulation have been reported especially in moderate drowsiness situations 69,71,91,94
or less vulnerable population to drowsiness. 72,91 In addition, there is still little research regarding the
usefulness of PERCLOS for detecting drowsiness in sleep disorders such as sleep apnea 55,99 or
aviation-related tasks. 64 To better understand the utility of PERCLOS, it will be necessary to
accumulate data regarding which populations and situations the PERCLOS is effective in.
Many studies measuring PERCLOS during PVT and the OSLER test have reported that
PERCLOS has the best accuracy among multiple measures. 47,52,58 However, studies that measured
indices during driving 59,68,71,81,101 or aviation-related tasks 64 have reported that other indices
exhibited superior performance for detecting drowsiness-related performance impairment
compared with PERCLOS (Table 2). These findings may have resulted from the following
technological limitations of PERCLOS: (1) inadequate detection of the eyes from the camera’s field of
view because of head movement; (2) occlusion of the face because of glasses, sunglasses, etc.; or (3)
inadequate measurement because of light reflection. 52,119 To overcome these limitations,
integrating several physiological and behavioral measures such as other eye and eyelid metrics, 51,120
HRV, 52,121 behavioral measures, 68 or contextual information 122,123 into PERCLOS may substitute for
the evaluation of drowsiness when PERCLOS is not detectable, contribute to improving the detection
accuracy of drowsiness-related performance deterioration by factors other than falling asleep, such
as inattention or distraction. 67,72 Integrating other indicators with PERCLOS may also contribute to
improvements of two issues: (1) decreasing of accuracy as the evaluation time becomes shorter not
only in PVT47,51 but also in driving tasks;71 (2) false positives or false negatives in the detection of
performance deterioration.52,58,67,71
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
Because of variation in the definition of PERCLOS and differences in sensing technology,
different devices may output different PERCLOS values. 71 The results also differ when PERCLOS and
%TEC are measured simultaneously.70 Therefore, standardization of PERCLOS will be necessary in
future research. In addition, many studies have used Optalert implementing JDS to conduct
validation studies in various environments. 114 Validation using a single device that implements a
PERCLOS-based method would facilitate comparisons between studies.
Although the development of real-time drowsiness measurements for drivers is particularly
advanced, drowsiness is also a critical issue in a range of other professions, including flight crews, 124
flight controllers, 54,125 astronauts, 126–128 and medical personnel. 129–131 Validation of the application
of drowsiness estimation in these professions may contribute to resolving drowsiness-related issues.
For example, PERCLOS has recently been implemented in devices such as network cameras. 132 Such
devices may enable early detection of drowsiness caused by work schedules 125,129 or work type 54 by
conducting periodic drowsiness measures of fixed-position staff 132 in offices, air traffic control units,
and nurses stations, in situations in which there is relatively little head movement, a minimal
influence of ambient light, and no need to hide the face with sunglasses. For this purpose, the
validation of drowsiness measurements during simulated tasks or real-world occupational scenarios
is needed. The ability to unobtrusively and objectively assess drowsiness will increase opportunities
for applying these measurements in the workplace, enable prompt countermeasures to drowsiness
in the workplace, and prevent occupational human errors caused by drowsiness.
Limitations
This review involved several limitations: First, because this was not a systematic review, the
possibility of potential bias cannot be ruled out. In addition, although this review extracted related
papers from several databases, including PubMed and Web of Science, some relevant studies may
not have been included. Second, because the focus of the present review was to discuss PERCLOS,
there may have been an insufficient discussion of drowsiness indices, such as blink duration or JDS,
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
which are other well-validated passive measures of drowsiness. Third, the current review did not
discuss the development of techniques or hardware for measuring PERCLOS with high levels of
accuracy (e.g., computer vision, machine learning, or sensors).
Conclusions
Many studies have shown that PERCLOS increases with sleep deprivation, after partial sleep
restriction, at nighttime, and by other drowsiness manipulations during vigilance tests, simulated
driving, and on-road driving. However, there are cases in which PERCLOS is not affected by
drowsiness manipulation, such as in moderate drowsiness conditions, in older adults, and during
aviation-related tasks. In addition, although PERCLOS is one of the most sensitive indices for
detecting drowsiness-related performance impairments during PVT or the OSLER test, other indices
exhibit superior performance compared with PERCLOS during simulated driving or aviation tasks.
Another potential limitation of PERCLOS is that slight differences in the definition of PERCLOS and
the presence of multiple measurement instruments may cause variations in PERCLOS values among
different studies. Standardization of PERCLOS or validation across multiple studies using a single
device to implement PERCLOS-based technology may be useful for clarifying the range of
populations, tasks, and situations for which it is an effective measure. Furthermore, PERCLOS alone
may not be sensitive for detecting drowsiness-related performance impairment by factors such as
inattention or distraction other than falling asleep. 67,72 The improvement of accuracy and validation
studies of PERCLOS as a drowsiness measure by combining it with other indicators may be useful for
overcoming its limitations. The development of PERCLOS-based technology will contribute to the
prevention of human error and accidents caused by drowsiness in various occupational fields.
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
Disclosure Statement
Financial Disclosure: This research was supported by the Japan Society for the Promotion of Science
(JSPS) KAKENHI Grant Numbers 17K19639 and 21K10357; the Ministry of Education, Culture, Sports,
Science, and Technology (MEXT) World Premier International Research Center Initiative (WPI)
program; and the Japan Agency for Medical Research and Development (AMED) (grant Number
JP21zf0127005).
Nonfinancial Disclosure: None
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
References
1. Dinges DF, Pack F, Williams K, et al. Cumulative sleepiness, mood disturbance and psychomotor
vigilance performance decrements during a week of sleep restricted to 4-5 hours per night.
Sleep. 1997;20(4):267-277. doi: 10.1093/sleep/20.4.267
2. Carskadon MA, Dement WC. Cumulative effects of sleep restriction on daytime sleepiness.
Psychophysiology. 1981;18(2):107-113. doi: 10.1111/j.1469-8986.1981.tb02921.x.
3. Van Dongen HPA, Maislin G, Mullington JM, Dinges DF. The cumulative cost of additional
wakefulness: dose-response effects on neurobehavioral functions and sleep physiology from
chronic sleep restriction and total sleep deprivation. Sleep. 2003;26(2):117-126. doi:
10.1093/sleep/26.2.117.
4. Belenky G, Wesensten NJ, Thorne DR, et al. Patterns of performance degradation and restoration
during sleep restriction and subsequent recovery: a sleep dose-response study. J Sleep Res.
2003;12(1):1-12. doi: 10.1046/j.1365-2869.2003.00337.x.
5. Åkerstedt T, Gillberg M. Subjective and objective sleepiness in the active individual. Int J Neurosci.
1990;52(1-2):29-37. doi: 10.3109/00207459008994241.
6. Cohen DA, Wang W, Wyatt JK, et al. Uncovering residual effects of chronic sleep loss on human
performance. Sci Transl Med. 2010;2(14):14ra3. doi: 10.1126/scitranslmed.3000458.
7. Zhou X, Ferguson SA, Matthews RW, et al. Mismatch between subjective alertness and objective
performance under sleep restriction is greatest during the biological night. J Sleep Res.
2012;21(1):40-49. doi: 10.1111/j.1365-2869.2011.00924.x.
8. Dijk DJ, Duffy JF, Czeisler CA. Circadian and sleep/wake dependent aspects of subjective alertness
and cognitive performance. J Sleep Res. 1992;1(2):112-117. doi: 10.1111/j.1365-
2869.1992.tb00021.x.
9. Zhou X, Ferguson SA, Matthews RW, et al. Sleep, wake and phase dependent changes in
neurobehavioral function under forced desynchrony. Sleep. 2011;34(7):931-941. doi:
10.5665/SLEEP.1130.
10. Lavie P. Ultrashort sleep-waking schedule. III. ‘Gates’ and ‘forbidden zones’ for sleep.
Electroencephalogr Clin Neurophysiol. 1986;63(5):414-425. doi: 10.1016/0013-4694(86)90123-
9.
11. Lavie P. Ultradian rhythms in arousal—the problem of masking. Chronobiol Int. 1989;6(1):21-28.
doi: 10.3109/07420528909059139.
12. Hayashi M, Sato K, Hori T. Ultradian rhythms in task performance, self-evaluation, and EEG
activity. Percept Mot Skills. 1994;79(2):791-800. doi: 10.2466/pms.1994.79.2.791.
13. Hull JT, Wright KP Jr, Czeisler CA. The influence of subjective alertness and motivation on human
performance independent of circadian and homeostatic regulation. J Biol Rhythms.
2003;18(4):329-338. doi: 10.1177/0748730403253584.
14. Mathew GM, Strayer SM, Bailey DS, et al. Changes in subjective motivation and effort during
sleep restriction moderate interindividual differences in attentional performance in healthy
young men. Nat Sci Sleep. 2021;13:1117-1136. doi: 10.2147/NSS.S294409.
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
15. Hayashi M, Minami S, Hori T. Masking effect of motivation on ultradian rhythm. Percept Mot
Skills. 1998;86(1):127-136. doi: 10.2466/pms.1998.86.1.127.
16. Jewett ME, Wyatt JK, Ritz-De Cecco A, Khalsa SB, Dijk DJ, Czeisler CA. Time course of sleep inertia
dissipation in human performance and alertness. J Sleep Res. 1999;8(1):1-8. doi:
10.1111/j.1365-2869.1999.00128.x.
17. McHill AW, Hull JT, Cohen DA, Wang W, Czeisler CA, Klerman EB. Chronic sleep restriction greatly
magnifies performance decrements immediately after awakening. Sleep. 2019;42(5). doi:
10.1093/sleep/zsz032.
18. Van Dongen HP, Price NJ, Mullington JM, Szuba MP, Kapoor SC, Dinges DF. Caffeine eliminates
psychomotor vigilance deficits from sleep inertia. Sleep. 2001;24(7):813-819. doi:
10.1093/sleep/24.7.813.
19. Bonnet MH, Arand DL. Arousal components which differentiate the MWT from the MSLT. Sleep.
2001;24(4):441-447. doi: 10.1093/sleep/24.4.441.
20. Cajochen C, Zeitzer JM, Czeisler CA, Dijk DJ. Dose-response relationship for light intensity and
ocular and electroencephalographic correlates of human alertness. Behav Brain Res.
2000;115(1):75-83. doi: 10.1016/s0166-4328(00)00236-9.
21. Goel N, Abe T, Braun ME, Dinges DF. Cognitive workload and sleep restriction interact to
influence sleep homeostatic responses. Sleep. 2014;37(11):1745-1756. doi: 10.5665/sleep.4164.
22. Baumann CR, Khatami R, Werth E, Bassetti CL. Hypocretin (orexin) deficiency predicts severe
objective excessive daytime sleepiness in narcolepsy with cataplexy. J Neurol Neurosurg
Psychiatry. 2006;77(3):402-404. doi: 10.1136/jnnp.2005.067207.
23. Schmidt C, Collette F, Leclercq Y, et al. Homeostatic sleep pressure and responses to sustained
attention in the suprachiasmatic area. Science. 2009;324(5926):516-519. doi:
10.1126/science.1167337.
24. Van Dongen HPA, Dinges DF. Investigating the interaction between the homeostatic and
circadian processes of sleep-wake regulation for the prediction of waking neurobehavioural
performance. J Sleep Res. 2003;12(3):181-187. doi: 10.1046/j.1365-2869.2003.00357.x:357 [pii].
25. Wang Z, Ma J, Miyoshi C, et al. Quantitative phosphoproteomic analysis of the molecular
substrates of sleep need. Nature. 2018;558(7710):435-439. doi: 10.1038/s41586-018-0218-8.
26. Funato H, Miyoshi C, Fujiyama T, et al. Forward-genetics analysis of sleep in randomly
mutagenized mice. Nature. 2016;539(7629):378-383. doi: 10.1038/nature20142.
27. Cheng P, Roehrs T, Roth T. Daytime sleepiness and alertness. In: Kryger M, Roth T, Goldstein CA,
Dement WC, eds. Principles and Practice of Sleep Medicine. 7th ed; 2022:35-45.
28. Cluydts R, De Valck E, Verstraeten E, Theys P. Daytime sleepiness and its evaluation. Sleep Med
Rev. 2002;6(2):83-96. doi: 10.1053/smrv.2002.0191.
29. Shen J, Barbera J, Shapiro CM. Distinguishing sleepiness and fatigue: focus on definition and
measurement. Sleep Med Rev. 2006;10(1):63-76. doi: 10.1016/j.smrv.2005.05.004.
30. Johns MW. Rethinking the assessment of sleepiness. Sleep Med Rev. 1998;2(1):3-15. doi:
10.1016/s1087-0792(98)90050-8.
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
31. Johns MW. A new perspective on sleepiness. Sleep Biol Rhythms. 2010;8(3):170-179. doi:
10.1111/j.1479-8425.2010.00450.x.
32. Sparrow AR, LaJambe CM, Van Dongen HPA. Drowsiness measures for commercial motor vehicle
operations. Accid Anal Prev. 2019;126:146-159. doi: 10.1016/j.aap.2018.04.020.
33. Curcio G, Casagrande M, Bertini M. Sleepiness: evaluating and quantifying methods. Int J
Psychophysiol. 2001;41(3):251-263. doi: 10.1016/s0167-8760(01)00138-6.
34. Carskadon MA, Dement WC. Effects of total sleep loss on sleep tendency. Percept Mot Skills.
1979;48(2):495-506. doi: 10.2466/pms.1979.48.2.495.
35. Williamson A, Lombardi DA, Folkard S, Stutts J, Courtney TK, Connor JL. The link between fatigue
and safety. Accid Anal Prev. 2011;43(2):498-515. doi: 10.1016/j.aap.2009.11.011.
36. Oken BS, Salinsky MC, Elsas SM. Vigilance, alertness, or sustained attention: physiological basis
and measurement. Clin Neurophysiol. 2006;117(9):1885-1901. doi:
10.1016/j.clinph.2006.01.017.
37. Lim J, Dinges DF. A meta-analysis of the impact of short-term sleep deprivation on cognitive
variables. Psychol Bull. 2010;136(3):375-389. doi: 10.1037/a0018883.
38. Basner M, Savitt A, Moore TM, et al. Development and validation of the cognition test battery for
spaceflight. Aerosp Med Hum Perform. 2015;86(11):942-952. doi: 10.3357/AMHP.4343.2015.
39. Dinges DF, Powell JW. Microcomputer analyses of performance on a portable, simple visual RT
task during sustained operations. Behav Res Methods Instrum Comput. 1985;17(6):652-655. doi:
10.3758/BF03200977.
40. Dorrian J, Rogers NL, Dinges DF. Psychomotor vigilance performance: neurocognitive assay
sensitive to sleep loss. In: Kushida CA, ed. Marcel Dekker; 2004:39-70. doi: 10.1201/b14100-5.
41. Lim J, Dinges DF. Sleep deprivation and vigilant attention. Ann N Y Acad Sci. 2008;1129:305-322.
doi: 10.1196/annals.1417.002.
42. Basner M, Dinges DF. Maximizing sensitivity of the psychomotor vigilance test (PVT) to sleep loss.
Sleep. 2011;34(5):581-591. doi: 10.1093/sleep/34.5.581.
43. Basner M, Dinges DF. An adaptive-duration version of the PVT accurately tracks changes in
psychomotor vigilance induced by sleep restriction. Sleep. 2012;35(2):193-202. doi:
10.5665/sleep.1620.
44. Basner M, Mollicone D, Dinges DF. Validity and sensitivity of a brief psychomotor vigilance test
(PVT-B) to total and partial sleep deprivation. Acta Astronaut. 2011;69(11-12):949-959. doi:
10.1016/j.actaastro.2011.07.015.
45. Wierwille WW, Wreggit SS, Kirn CL, Ellsworth LA, Fairbanks RJ. Research on Vehicle-Based Driver
Status/Performance Monitoring; Development, Validation, and Refinement of Algorithms for
Detection of Driver Drowsiness. Final Report. National Highway Traffic Safety Administration
Final Report. 1994.
46. Wierwille WW, Ellsworth LA. Evaluation of driver drowsiness by trained raters. Accid Anal Prev.
1994;26(5):571-581. doi: 10.1016/0001-4575(94)90019-1.
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
47. Dinges DF, Mallis M, Maislin G, Powell JW. Evaluation of techniques for ocular measurement as
an index of fatigue and as the basis for alertness management. United States Department of
Transportation, National Highway Traffic Safety Administration. 1998.
48. Abe T, Mollicone D, Basner M, Dinges DF. Sleepiness and safety: where biology needs
technology. Sleep Biol Rhythms. 2014;12(2):74-84. doi: 10.1111/sbr.12067.
49. Cori JM, Anderson C, Shekari Soleimanloo S, Jackson ML, Howard ME. Narrative review: do
spontaneous eye blink parameters provide a useful assessment of state drowsiness? Sleep Med
Rev. 2019;45:95-104. doi: 10.1016/j.smrv.2019.03.004.
50. Cai AWT, Manousakis JE, Lo TYT, Horne JA, Howard ME, Anderson C. I think I’m sleepy, therefore
I am – Awareness of sleepiness while driving: A systematic review. Sleep Med Rev.
2021;60:101533. doi: 10.1016/j.smrv.2021.101533.
51. Abe T, Mishima K, Kitamura S, et al. Tracking intermediate performance of vigilant attention
using multiple eye metrics. Sleep. 2020;43(3). doi: 10.1093/sleep/zsz219.
52. Chua ECP, Tan WQ, Yeo SC, et al. Heart rate variability can be used to estimate sleepiness-related
decrements in psychomotor vigilance during total sleep deprivation. Sleep. 2012;35(3):325-334.
doi: 10.5665/sleep.1688.
53. Merat N, Jamson AH. The effect of three low-cost engineering treatments on driver fatigue: A
driving simulator study. Accid Anal Prev. 2013;50:8-15. doi: 10.1016/j.aap.2012.09.017.
54. Zhang J, Chen Z, Liu W, Ding P, Wu Q. A field study of work type influence on air traffic
controllers’ fatigue based on data-driven PERCLOS detection. Int J Environ Res Public Health.
2021;18(22). doi: 10.3390/ijerph182211937.
55. Kratzel L, Glos M, Veauthier C, et al. Video-based sleep detection using ocular signals under the
standard conditions of the maintenance of wakefulness test in patients with sleep disorders.
Physiol Meas. 2021;42(1):014004. doi: 10.1088/1361-6579/abdb7e.
56. Salvati L, d’Amore M, Fiorentino A, Pellegrino A, Sena P, Villecco F. On-road detection of driver
fatigue and drowsiness during medium-distance journeys. Entropy (Basel). 2021;23(2). doi:
10.3390/e23020135.
57. Murata A, Doi T, Karwowski W. Sensitivity of PERCLOS70 to drowsiness level: effectiveness of
PERCLOS70 to prevent crashes caused by drowsiness. IEEE Access. 2022;10:70806-70814. doi:
10.1109/ACCESS.2022.3187995.
58. Abe T, Nonomura T, Komada Y, et al. Detecting deteriorated vigilance using percentage of eyelid
closure time during behavioral maintenance of wakefulness tests. Int J Psychophysiol.
2011;82(3):269-274. doi: 10.1016/j.ijpsycho.2011.09.012.
59. Puspasari MA, Iridiastadi H, Sutalaksana IZ, Sjafruddin A. Ocular indicators as fatigue detection
instruments for Indonesian drivers. Ind Eng Manag Syst. 2019;18(4):748-760. doi:
10.7232/iems.2019.18.4.748.
60. Soukupová T, Cech J. Real-time eye blink detection using facial landmarks. 21st Computer Vision
Winter Workshop. 2016:3-5. http://vision.fe.uni-lj.si/cvww2016/proceedings/papers/05.pdf
61. Cheng Q, Wang W, Jiang X, Hou S, Qin Y. Assessment of driver mental fatigue using facial
landmarks. IEEE Access. 2019;7:150423-150434. doi: 10.1109/ACCESS.2019.2947692.
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
62. Chua ECP, Yeo SC, Lee ITG, et al. Individual differences in physiologic measures are stable across
repeated exposures to total sleep deprivation. Physiol Rep. 2014;2(9). doi:
10.14814/phy2.12129.
63. Thropp JE, Scallon JFV, Buza P. PERCLOS as an indicator of slow-onset hypoxia in aviation. Aerosp
Med Hum Perform. 2018;89(8):700-707. doi: 10.3357/AMHP.5059.2018.
64. McKinley RA, McIntire LK, Schmidt R, Repperger DW, Caldwell JA. Evaluation of eye metrics as a
detector of fatigue. Hum Factors. 2011;53(4):403-414. doi: 10.1177/0018720811411297.
65. Alvaro PK, Jackson ML, Berlowitz DJ, Swann P, Howard ME. Prolonged eyelid closure episodes
during sleep deprivation in professional drivers. J Clin Sleep Med. 2016;12(8):1099-1103. doi:
10.5664/jcsm.6044.
66. Grace R, Byrne VE, Bierman DM, et al. A drowsy driver detection system for heavy vehicles.
ieeexplore.ieee.org. In: 17th DASC. AIAA/IEEE/SAE. Proceedings. Digital Avionics Syst
Conference:vol. 2; 1998:I36/1-I36 vol.2. doi: 10.1109/DASC.1998.739878 (Cat. No. 98CH36267).
67. Jackson ML, Raj S, Croft RJ, et al. Slow eyelid closure as a measure of driver drowsiness and its
relationship to performance. Traffic Inj Prev. 2016;17(3):251-257. doi:
10.1080/15389588.2015.1055327.
68. McDonald AD, Lee JD, Schwarz C, Brown TL. A contextual and temporal algorithm for driver
drowsiness detection. Accid Anal Prev. 2018;113:25-37. doi: 10.1016/j.aap.2018.01.005.
69. Caponecchia C, Williamson A. Drowsiness and driving performance on commuter trips. Safety
Res. 2018;66:179-186. doi: 10.1016/j.jsr.2018.07.003.
70. Jackson ML, Kennedy GA, Clarke C, et al. The utility of automated measures of ocular metrics for
detecting driver drowsiness during extended wakefulness. Accid Anal Prev. 2016;87:127-133.
doi: 10.1016/j.aap.2015.11.033.
71. Golz M, Sommer D, Trutschel U, Sirois B, Edwards D. Evaluation of fatigue monitoring
technologies. Somnologie. 2010;14(3):187-199. doi: 10.1007/s11818-010-0482-9.
72. Cai AWT, Manousakis JE, Singh B, et al. On-road driving impairment following sleep deprivation
differs according to age. Sci Rep. 2021;11(1):21561. doi: 10.1038/s41598-021-99133-y.
73. Anderson C, Chang AM, Sullivan JP, Ronda JM, Czeisler CA. Assessment of drowsiness based on
ocular parameters detected by infrared reflectance oculography. J Clin Sleep Med.
2013;9(9):907-920:920A-920B. doi: 10.5664/jcsm.2992.
74. Ftouni S, Rahman SA, Crowley KE, Anderson C, Rajaratnam SMW, Lockley SW. Temporal
dynamics of ocular indicators of sleepiness across sleep restriction. J Biol Rhythms.
2013;28(6):412-424. doi: 10.1177/0748730413512257.
75. Liang Y, Horrey WJ, Howard ME, et al. Prediction of drowsiness events in night shift workers
during morning driving. Accid Anal Prev. 2019;126:105-114. doi: 10.1016/j.aap.2017.11.004.
76. Ong JL, Asplund CL, Chia TTY, Chee MWL. Now you hear me, now you don’t: eyelid closures as an
indicator of auditory task disengagement. Sleep. 2013;36(12):1867-1874. doi:
10.5665/sleep.3218.
77. Ong JL, Kong D, Chia TTY, Tandi J, Thomas Yeo BT, Chee MWL. Co-activated yet disconnected-
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
Neural correlates of eye closures when trying to stay awake. Neuroimage. 2015;118:553-562.
doi: 10.1016/j.neuroimage.2015.03.085.
78. Wang C, Ong JL, Patanaik A, Zhou J, Chee MWL. Spontaneous eyelid closures link vigilance
fluctuation with fMRI dynamic connectivity states. Proc Natl Acad Sci U S A. 2016;113(34):9653-
9658. doi: 10.1073/pnas.1523980113.
79. De Havas JA, Parimal S, Soon CS, Chee MWL. Sleep deprivation reduces default mode network
connectivity and anti-correlation during rest and task performance. Neuroimage.
2012;59(2):1745-1751. doi: 10.1016/j.neuroimage.2011.08.026.
80. Bennett LS, Stradling JR, Davies RJ. A behavioural test to assess daytime sleepiness in obstructive
sleep apnoea. J Sleep Res. 1997;6(2):142-145. doi: 10.1046/j.1365-2869.1997.00039.x.
81. Wilkinson VE, Jackson ML, Westlake J, et al. Assessing the validity of eyelid parameters to detect
impairment due to benzodiazepines. Hum Psychopharmacol. 2020;35(2):e2723. doi:
10.1002/hup.2723.
82. Ftouni S, Sletten TL, Nicholas CL, Kennaway DJ, Lockley SW, Rajaratnam SMW. Ocular measures
of sleepiness are increased in night shift workers undergoing a simulated night shift near the
peak time of the 6-sulfatoxymelatonin rhythm. J Clin Sleep Med. 2015;11(10):1131-1141. doi:
10.5664/jcsm.5086.
83. Johns MW, Tucker A, Chapman R, Crowley K, Michael N. Monitoring eye and eyelid movements
by infrared reflectance oculography to measure drowsiness in drivers. Somnologie.
2007;11(4):234-242. doi: 10.1007/s11818-007-0311-y.
84. Wilkinson VE, Jackson ML, Westlake J, et al. The accuracy of eyelid movement parameters for
drowsiness detection. J Clin Sleep Med. 2013;9(12):1315-1324. doi: 10.5664/jcsm.3278.
85. Dennis LE, Wohl RJ, Selame LA, Goel N. Healthy adults display long-term trait-like
neurobehavioral resilience and vulnerability to sleep loss. Sci Rep. 2017;7(1):14889. doi:
10.1038/s41598-017-14006-7.
86. Van Dongen HPA, Baynard MD, Maislin G, Dinges DF. Systematic interindividual differences in
neurobehavioral impairment from sleep loss: evidence of trait-like differential vulnerability.
Sleep. 2004;27(3):423-433. doi: 10.1093/sleep/27.3.423.
87. Yamazaki EM, Goel N. Robust stability of trait-like vulnerability or resilience to common types of
sleep deprivation in a large sample of adults. Sleep. 2020;43(6). doi: 10.1093/sleep/zsz292.
88. Balkin TJ, Horrey WJ, Graeber RC, Czeisler CA, Dinges DF. The challenges and opportunities of
technological approaches to fatigue management. Accid Anal Prev. 2011;43(2):565-572. doi:
10.1016/j.aap.2009.12.006.
89. Mortazavi A, Eskandarian A, Sayed RA. Effect of drowsiness on driving performance variables of
commercial vehicle drivers. Int.J Automot Technol. 2009;10(3):391-404. doi: 10.1007/s12239-
009-0045-x.
90. Wörle J, Metz B, Steinborn MB, Huestegge L, Baumann M. Differential effects of driver sleepiness
and sleep inertia on driving behavior. Transp Res F. 2021;82:111-120. doi:
10.1016/j.trf.2021.08.001.
91. Arefnezhad S, Eichberger A, Koglbauer IV. Effects of automation and fatigue on drivers from
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
various age groups. Safety. 2022;8(2):30. doi: 10.3390/safety8020030.
92. Wang X, Xu C. Driver drowsiness detection based on non-intrusive metrics considering individual
specifics. Accid Anal Prev. 2016;95(B):350-357. doi: 10.1016/j.aap.2015.09.002.
93. Ftouni S, Sletten TL, Howard M, et al. Objective and subjective measures of sleepiness, and their
associations with on-road driving events in shift workers. J Sleep Res. 2013;22(1):58-69. doi:
10.1111/j.1365-2869.2012.01038.x.
94. Ahlström C, Zemblys R, Jansson H, Forsberg C, Karlsson J, Anund A. Effects of partially automated
driving on the development of driver sleepiness. Accid Anal Prev. 2021;153:106058. doi:
10.1016/j.aap.2021.106058.
95. Lee ML, Howard ME, Horrey WJ, et al. High risk of near-crash driving events following night-shift
work. Proc Natl Acad Sci U S A. 2016;113(1):176-181. doi: 10.1073/pnas.1510383112.
96. Smith A, McDonald AD, Sasangohar F. The impact of commutes, work schedules, and sleep on
near-crashes during nurses’ post shift-work commutes: A naturalistic driving study. IISE Trans
Occup Ergon Hum Factors. 2021;9(1):13-22. doi: 10.1080/24725838.2021.1945708.
97. Scott LD, Hwang WT, Rogers AE, Nysse T, Dean GE, Dinges DF. The relationship between nurse
work schedules, sleep duration, and drowsy driving. Sleep. 2007;30(12):1801-1807. doi:
10.1093/sleep/30.12.1801.
98. Anderson C, Wales AWJ, Horne JA. PVT lapses differ according to eyes open, closed, or looking
away. Sleep. 2010;33(2):197-204. doi: 10.1093/sleep/33.2.197.
99. Cori JM, Turner S, Westlake J, et al. Eye blink parameters to indicate drowsiness during
naturalistic driving in participants with obstructive sleep apnea: A pilot study. Sleep Health.
2021;7(5):644-651. doi: 10.1016/j.sleh.2021.01.009.
100. Dingus TA, Hardee HL, Wierwille WW. Development of models for on-board detection of driver
impairment. Accid Anal Prev. 1987;19(4):271-283. doi: 10.1016/0001-4575(87)90062-5.
101. McDonald AD, Lee JD, Schwarz C, Brown TL. Steering in a random forest: ensemble learning for
detecting drowsiness-related lane departures. Hum Factors. 2014;56(5):986-998. doi:
10.1177/0018720813515272.
102. Hanowski R, Bowman DS, Alden A, Wierwille W, Carroll RJ, PERCLOS+: Development of a Robust
Field Measure of Driver Drowsiness. 15th World Congress on Intelligent Transport Systems and
ITS America's 2008 Annual Meeting, 2008, 12p
103. Hanowski RJ, Bowman D, Alden A, Wierwille WW, Carroll R. PERCLOS+:Moving beyond single-
metric drowsiness monitors. SAE Technical Paper Series. 2008. doi: 10.4271/2008-01-2692.
104. Hertig-Godeschalk A, Skorucak J, Malafeev A, Achermann P, Mathis J, Schreier DR. Microsleep
episodes in the borderland between wakefulness and sleep. Sleep. 2020;43(1). doi:
10.1093/sleep/zsz163.
105. Devos H, Alissa N, Lynch S, Sadeghi M, Akinwuntan AE, Siengsukon C. Real-time assessment of
daytime sleepiness in drivers with multiple sclerosis. Mult Scler Relat Disord. 2021;47:102607.
doi: 10.1016/j.msard.2020.102607.
106. Funke ME, Warm JS, Matthews G, et al. The neuroergonomics of vigilance. Hum Factors.
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
2017;59(1):62-75. doi: 10.1177/0018720816683121.
107. Mulhall MD, Cori J, Sletten TL, et al. A pre-drive ocular assessment predicts alertness and
driving impairment: A naturalistic driving study in shift workers. Accid Anal Prev.
2020;135:105386. doi: 10.1016/j.aap.2019.105386.
108. Dinges DF, Maislin G, Brewster RM, Krueger GP, Carroll RJ. Pilot test of fatigue management
technologies. Transp Res Rec. 2005;1922(1):175-182. doi: 10.1177/0361198105192200122.
109. Johns MW, Chapman R, Crowley K, Tucker A. A new method for assessing the risks of
drowsiness while driving. Somnologie. 2008;12(1):66-74. doi: 10.1007/s11818-008-0330-3.
110. Aidman E, Johnson K, Paech GM, et al. Caffeine reduces the impact of drowsiness on driving
errors. Transp Res F. 2018;54:236-247. doi: 10.1016/j.trf.2018.01.008.
111. Michael N, Johns M, Owen C, Patterson J. Effects of caffeine on alertness as measured by
infrared reflectance oculography. Psychopharmacology. 2008;200(2):255-260. doi:
10.1007/s00213-008-1202-z.
112. Aidman E, Chadunow C, Johnson K, Reece J. Real-time driver drowsiness feedback improves
driver alertness and self-reported driving performance. Accid Anal Prev. 2015;81:8-13. doi:
10.1016/j.aap.2015.03.041.
113. Shekari Soleimanloo S, Wilkinson VE, Cori JM, et al. Eye-blink parameters detect on-road track-
driving impairment following severe sleep deprivation. J Clin Sleep Med. 2019;15(9):1271-1284.
doi: 10.5664/jcsm.7918.
114. Dawson D, Searle AK, Paterson JL. Look before you (s)leep: evaluating the use of fatigue
detection technologies within a fatigue risk management system for the road transport
industry. Sleep Med Rev. 2014;18(2):141-152. doi: 10.1016/j.smrv.2013.03.003.
115. Åkerstedt T, Ingre M, Kecklund G, et al. Reaction of sleepiness indicators to partial sleep
deprivation, time of day and time on task in a driving simulator—the DROWSI project. J Sleep
Res. 2010;19(2):298-309. doi: 10.1111/j.1365-2869.2009.00796.x.
116. Ingre M, Åkerstedt T, Peters B, Anund A, Kecklund G. Subjective sleepiness, simulated driving
performance and blink duration: examining individual differences. J Sleep Res. 2006;15(1):47-
53. doi: 10.1111/j.1365-2869.2006.00504.x.
117. Hallvig D, Anund A, Fors C, Kecklund G, Åkerstedt T. Real driving at night—predicting lane
departures from physiological and subjective sleepiness. Biol Psychol. 2014;101:18-23. doi:
10.1016/j.biopsycho.2014.07.001.
118. Schwarz JFA, Ingre M, Fors C, et al. In-car countermeasures open window and music revisited
on the real road: popular but hardly effective against driver sleepiness. J Sleep Res.
2012;21(5):595-599. doi: 10.1111/j.1365-2869.2012.01009.x.
119. Abbas Q, Alsheddy A. A methodological review on prediction of multi-stage hypovigilance
detection systems using multimodal features. IEEE Access. 2021;9:47530-47564. doi:
10.1109/ACCESS.2021.3068343.
120. François C, Hoyoux T, Langohr T, Wertz J, Verly JG. Tests of a new drowsiness characterization
and monitoring system based on ocular parameters. Int J Environ Res Public Health.
2016;13(2):174. doi: 10.3390/ijerph13020174.
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
121. Henelius A, Sallinen M, Huotilainen M, Müller K, Virkkala J, Puolamäki K. Heart rate variability
for evaluating vigilant attention in partial chronic sleep restriction. Sleep. 2014;37(7):1257-
1267. doi: 10.5665/sleep.3850.
122. Ji Q, Lan P, Looney C. A probabilistic framework for modeling and real-time monitoring human
fatigue. IEEE Trans Syst Man Cybern A. 2006;36(5):862-875. doi: 10.1109/TSMCA.2005.855922.
123. Yang G, Lin Y, Bhattacharya P. A driver fatigue recognition model based on information fusion
and dynamic Bayesian network. Information Sciences. 2010;180(10):1942-1954. doi:
10.1016/j.ins.2010.01.011.
124. Neri DF, Oyung RL, Colletti LM, Mallis MM, Tam PY, Dinges DF. Controlled breaks as a fatigue
countermeasure on the flight deck. Aviat Space Environ Med. 2002;73(7):654-664.
https://www.ncbi.nlm.nih.gov/pubmed/12137101.
125. Mizuno K, Matsumoto A, Aiba T, et al. Sleep patterns among shift-working flight controllers of
the International Space Station: an observational study on the JAXA Flight Control Team. J
Physiol Anthropol. 2016;35(1):19. doi: 10.1186/s40101-016-0108-4.
126. Jones CW, Basner M, Mollicone DJ, Mott CM, Dinges DF. Sleep deficiency in spaceflight is
associated with degraded neurobehavioral functions and elevated stress in astronauts on six-
month missions aboard the International Space Station. Sleep. 2022;45(3). doi:
10.1093/sleep/zsac006.
127. Dijk DJ, Neri DF, Wyatt JK, et al. Sleep, performance, circadian rhythms, and light-dark cycles
during two space shuttle flights. Am J Physiol Regul Integr Comp Physiol. 2001;281(5):R1647-
R1664. doi: 10.1152/ajpregu.2001.281.5.R1647.
128. Basner M, Dinges DF, Mollicone D, et al. Mars 520-d mission simulation reveals protracted crew
hypokinesis and alterations of sleep duration and timing. Proc Natl Acad Sci U S A.
2013;110(7):2635-2640. doi: 10.1073/pnas.1212646110.
129. Basner M, Asch DA, Shea JA, et al. Sleep and alertness in a duty-hour flexibility trial in internal
medicine. N Engl J Med. 2019;380(10):915-923. doi: 10.1056/NEJMoa1810641.
130. Basner M, Dinges DF, Shea JA, et al. Sleep and alertness in medical interns and residents: an
observational study on the role of extended shifts. Sleep. 2017;40(4). doi:
10.1093/sleep/zsx027.
131. Gander P, Millar M, Webster C, Merry A. Sleep loss and performance of anaesthesia trainees
and specialists. Chronobiol Int. 2008;25(6):1077-1091. doi: 10.1080/07420520802551428.
132. Nie B, Huang X, Chen Y, Li A, Zhang R, Huang J. Experimental study on visual detection for
fatigue of fixed-position staff. Appl Ergon. 2017;65:1-11. doi: 10.1016/j.apergo.2017.05.010.
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
Table 1. Summary of the effects of drowsiness manipulation on PERCLOS
Reference
Device
Task
Participants
(N, type, age range)
Drowsiness manipulation
Results
Fixation, PVT, OSLER test
Dinges et al., 1998 47
CCTV Camera
PVT
N = 10; healthy adults; 21–35
years
42 h of sleep deprivation
Changes corresponded to
increased PVT lapses in the
42-h period
Abe et al., 2011 58
EMR-9a
OSLER test
N = 9; healthy adults; 19–30 years
One night of partial sleep
deprivation (4 h sleep)
Changes corresponded to
increased missed response
after partial sleep
deprivation
McKinley et al., 2011
64
EC6 b
PVT
N = 10; healthy adults; 18–42
years
Experimental testing began after
14 h awake and continued until
28 h of sleep deprivation
No change
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
Chua et al., 2012 52
ISCAN eye-
tracker c
PVT
N = 24; healthy adults; 25.9 ± 2.8
years
40 h of sleep deprivation
Changes corresponded to
the profile of PVT
performance
Anderson et al., 2013
73
Optalert d
Looking ahead
N = 28; healthy adults; 18–34
years
30 h of sleep deprivation
Increased after 26 h of
wakefulness compared
with the first 16 h of
wakefulness
Ftouni et al., 2013 74
Optalert d
Auditory PVT
N = 10; healthy adults; 20–25
years
40 h of sleep deprivation
Increased at 24 to 25 h of
wakefulness compared
with the first 16 h after
awakening.
Chua et al., 2014 62
ISCAN eye-
tracker c
PVT
N = 12; healthy adults; 22–30
years
26 h of sleep deprivation
Increased after usual
bedtime
Ftouni et al., 2015 82
Optalert d
Auditory PVT
N = 22; night shift workers; 33.4 ±
11.8 years
Tested groups (1) within (N = 14)
or (2) outside (N = 8) the
acrophase (± 3 h from the peak)
of 6-sulphatoxymelatonin
(aMT6s)
Increased in group (1)
compared with group (2)
Jackson et al., 2016
70
Copilot 66
PVT
N = 22; healthy adults; 18–26
years
(1) One night of sleep
deprivation; (2) a normal night of
sleep
No change
Jackson et al., 2016
Optalert d
PVT
N = 22; healthy adults; 18–26
(1) One night of sleep
deprivation; (2) a normal night of
Increased in condition (1)
compared with condition
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
70
years
sleep
(2)
Abe et al., 2020 51
EMR-9 a
PVT
N = 16; healthy adults; 20–49
years
38 h of sleep deprivation
Increased after 18 h of
awake, peaked at 22 h
Wilkinson et al., 2020
81
Optalert d
PVT
N = 15; healthy adults; 37.6 ± 11.6
years
(1) Benzodiazepine
administration; (2) placebo
Increased after (1)
compared with (2)
Zhang et al., 2021 54
A front facing
laptop
camera
A simple reaction
task
N = 32; air traffic controllers at
terminal control unit; 23–36 years
Four shift zones: (1) 8:00–12:00;
(2) 12:00–18:00; (3)18:00–24:00;
(4) 24:00–8:00
Increased in shift (3) and
(4) compared with shift (1)
and (2)
Zhang et al., 2021 54
A front facing
laptop
camera
A simple reaction
task
N = 35; air traffic controllers at air
control unit; 23–38 years
Four shift zones: (1) 8:00–12:00;
(2) 12:00–18:00; (3)18:00–24:00;
(4) 24:00–8:00
No difference
Simulated driving
Dingus et al., 1987 100
Unobtrusive
camera/monitor
system
Fixed based;
nighttime
interstate driving
N = 6; drivers; over 21 years
(1) Driving at 19 h after
participants’ normal wake-up
time; (2) driving in the early
evening after a normal night’s
sleep
Changes corresponded to
the lane position measures
Mortazavi et al.,
2009 89
Head-mounted
eye tracking
system
Fixed based;
monotonous
highway driving
N = 13; professional truck drivers;
23–55 years
(1) A night session after 18–19 h
of wakefulness; (2) morning
session after at least 8 h of sleep
Increased in the night
session (1) compared with
the morning session (2)
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
Golz et al., 2010 71
Three
anonymous
devices
Fixed based;
monotonous
motorway
driving
N = 16; healthy adults; 19–29
years
Overnight driving sessions from
11:30 PM to 8:30 AM
Corresponded changes
with KSS and SDL that
progressively increased
throughout the night and
peaked at 6 AM.
Merat & Jamson,
2013 53
FaceLAB e
High fidelity;
motorway
driving
N = 17; shift workers; 31.4 ± 5.4
years
(1) After a night shift; (2) after a
normal night’s sleep
Increased in (1) compared
with (2)
Merat & Jamson,
2013 53
FaceLAB e
High fidelity;
motorway
driving
N = 16; older adults; 53.2 ± 5.5
years
(1) Afternoon after consuming a
large lunch; (2) baseline trial
during the morning
Increased in (1) compared
with (2)
Alvaro et al., 2016 65
Digital video
recordings using
infrared light
Fixed based;
monotonous
nighttime
highway driving
N = 20; professional drivers; 41.9
± 8.3 years
24 h of sleep deprivation
Increased during sleep
deprivation
Jackson et al., 2016 70
Copilot 66
Fixed based;
monotonous
nighttime
highway driving
N = 22; healthy adults; 18–26
years
(1) One night of sleep
deprivation; (2) after a normal
night of sleep
No change
Jackson et al., 2016 70
Optalert d
Fixed based;
monotonous
nighttime
highway driving
N = 22; healthy adults; 18–26
years
(1) One night of sleep
deprivation; (2) after a normal
night of sleep
Increased in (1) compared
with (2)
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
Jackson et al., 2016 67
Copilot 66
Fixed based;
monotonous
nighttime
highway driving
N = 12; healthy professional
drivers; 23–62 years
(1) After 24 h of sleep
deprivation; (2) after a normal
night of sleep
Increased in (1) compared
with (2)
Wang and Xu, 2016 92
SmartEye Pro f
High fidelity
motion-based
highway daytime
driving
N = 16; Night shift workers; 24–40
years
Driving after having worked an 8
h night shift
Increased with the KSS
measured drowsiness
levels
Caponecchia &
Williamson, 2018 69
SmartEye Pro f
Fixed based;
arterial road and
highway driving
N = 41; healthy adults; Group 1:
37.0 ± 8.3 years; Group 2: 42.1 ±
6.3 years; Group 3: 38.4 ± 5.9
years
Different level of sleep
deprivation (Group 1: no sleep
deprivation; Group 2: 2 h
deprivation; Group 3: 4 h
deprivation) and time of day
(morning and evening)
No change
Puspasari et al., 2019
59
EyeLink IIg
Fixed based;
highway driving
N = 13; healthy adults; 30.2 ± 5.0
years
Sleep duration was divided into
two levels: (1) 4 h or (2) 8 h
Increased in (1) compared
with (2)
Wilkinson et al., 2020
81
Optalert d
Fixed based;
monotonous
nighttime
highway driving
N = 15; healthy adults; 37.6 ± 11.6
years
(1) Benzodiazepine
administration; (2) placebo
Increased in (1) compared
with (2)
Wörle et al., 2021 90
SmartEye Pro f
High fidelity;
monotonous
manual driving
after take-over
N = 61; regular drivers; 38.1 ± 11.9
years
KSS determined “alert” (KSS ≤ 4)
or "sleepy" (KSS ≥ 7); EEG defined
“after sleep”
Increased after sleep
relative to the alert state.
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
from automated
driving
Murata et al., 2022 57
Drowsimeter
R100 h
Fixed based;
driving along a
straight road
with following a
leading car
N = 13; healthy adults; 21–25
years
(1) Drowsy state: one night of
staying up all night; (2) aroused
state: sufficient sleep
Increased in (1) compared
with (2)
Arefnezhad et al.,
2022 91
N/A
Fixed based with
four bass shakers
for generating
the vibration in
the car;
motorway
automated and
manual driving
N = 89; drivers; 20–85 years
(1) Fatigued conditions (16 h
awake or 4 h sleep the previous
night); (2) rested condition
Differences between the
automated and manual
mode were greater in (1)
compared with (2)
On-road driving
Ftouni et al., 2013 74
Optalert d
Participants'
commute to or
from work
N = 27; rotating and permanent
night shift-working nurses; 41.6 ±
12.5 years
Driving to and from night shifts
Increased after night shift
compared with that before
night shift
Lee et al., 2016 95
Optalert d
2-h daytime
driving on a
closed driving
track
N = 16; night shift workers; 18–65
years
(1) Post-night shift driving
following night shift work; (2)
driving after a night of at least 5 h
of sleep with no night shift work
No change
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
Ahlström et al., 2021
94
Smart Eye
System f
Automated
driving of a 90-
km section on a
dual-lane
motorway with
real traffic
N = 89; healthy drivers
(1) Nighttime (1:00 or 3:00); (2)
daytime (15:00 or 17:00)
Increased at nighttime
Cai et al., 2021 72
Seeing
Machines' DMS
e
2 h driving
around a closed
driving track
N = 16; young adults; 21–33 years
(1) After 29 h of sleep
deprivation; (2) after 8 h sleep
Increased after sleep
deprivation
Cai et al., 2021 72
Seeing
Machines' DMS
e
2 h driving
around a closed
driving track
N = 17; older adults: 50–65 years
(1) After 29 h of sleep
deprivation; (2) after 8 h sleep
No change
Cori et al., 202199
Optalert d
A week of
regular
naturalistic
driving
N = 30; Obstructive sleep apnea
(N = 15); 46.5 ± 7.3 years; healthy
controls (N = 15); 48.7 ± 7.2 years
Obstructive sleep apnea patients
or healthy controls
No difference
Aviation-related task
McKinley et al., 2011
64
EC6 b
Target
acquisition task
N = 10; healthy adults; 18–42
years
Experimental testing began after
14 h of wakefulness and
continued until 28 h of sleep
deprivation
No change
McKinley et al., 2011
EC6 b
Unmanned aerial
N = 10; healthy adults; 18–42
Experimental testing began after
14 h of wakefulness and
No change
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
64
vehicle task
years
continued until 28 h of sleep
deprivation
Optalert measured %TEC (percentage of time with eyes closed); CCTV: closed-circuit television; DMS: driver monitoring system; KSS: Karolinska Sleepiness
Scale; OSLER: Oxford Sleep Resistance; PVT: Psychomotor Vigilance Test; SDL: standard deviation of lateral position in lane; a nac Image Technology Inc., Tokyo,
Japan; b EyeCom, Inc., Reno, Nevada; c ISCAN, Inc., Woburn, MA; d Sleep Diagnostics Pty Ltd., Melbourne, Australia; e Seeing Machines, Canberra, Australia; f Smart Eye AB,
Gothenburg, Sweden; g SR-Research, Osgoode, Ontario, Canada; h Phasya S.A., Seraing, Belgium
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
Table 2. Summary of comparisons of accuracy between PERCLOS and other indicators in drowsiness detection
Reference
Drowsiness manipulation
Calculation method for accuracy
The best index for
identifying drowsiness
events
Fixation, PVT, OSLER test
Dinges et al., 1998 47
42 h of sleep deprivation
Pearson’s or Spearman’s rank correlations: (1) PVT lapse frequency;
(2) cumulative lapse duration time
(1–2) PERCLOS
Abe et al., 2011 58
One night of partial sleep
deprivation (4 h sleep)
AUC for discriminating (1) between the EP0 (no missed response)
epoch and EP1–6 (at least one missed response) epoch and (2)
between the EP0–2 (two consecutive missed responses) epoch and
EP3–6 (three consecutive missed responses) epoch.
(1–2) PERCLOS
Chua et al., 2012 52
40 h of sleep deprivation
(1) Pearson’s correlation analysis; (2) AUC to identify a threshold
increase (>25%, > 50%, or > 75%) relative to baseline in PVT lapses.
(1–2) PERCLOS a
Anderson et al., 2013
73,b
30 h of sleep deprivation
AUC to identify a threshold increase ((1) > 50%; (2) > 75%) relative to
baseline in PVT lapses (Recordings were obtained while participant
looked directly ahead); (3) AUC to identify KSS 6 or above
(1–3) JDS
Ftouni et al., 2013 74
40 h of sleep deprivation
AUC to identify a threshold increase in auditory PVT lapses ((1) > 25 %,
(2) > 50 %, (3) > 75%) relative to the first 16 h of wakefulness
(1–3) PosAVR
Wilkinson et al., 2013
84,c
8 h sleep or one night of 4 h
sleep restriction
AUC of missed signals: (1) ≥ 4 consecutive missed signals; (2) ≥ 4 total
missed signals in OSLER test
(1–2) IED
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
Chua et al., 2014 62
26 h of sleep deprivation
Intraclass correlation coefficient to assess the reproducibility of
individual differences between two visits of 26 h of sustained
wakefulness during PVT
PERCLOS
Jackson et al., 2016 70
One night of sleep deprivation
or after a normal night of sleep
Spearman’s rank correlation: (1) median RT; (2) lapse during PVT
(1–2) %TEC
Wilkinson et al., 2020
81
Benzodiazepine administration
or placebo
(1) AUC for detecting one or more PVT lapses; (2) AUC for detecting
two or more PVT lapses in 1-min bin
(1) JDS and (2) %TEC in the
placebo condition; (1) JDS
and (2) IED in the BZ
condition
Abe et al., 2020 51
38 h of sleep deprivation
Intraclass correlation coefficients between probabilities of PVT
response (RT ≥ 300 ms) and a new PERCLOS-based algorithm (PVT-E)
or PERCLOS
PVT-E
Simulated driving
Golz et al., 2010 71
Overnight driving sessions from
11:30 PM to 8:30 AM
Test error of nonlinear discrimination analysis for (1) KSS and (2)
standard deviation of lateral position in lane
(1–2) EEG/EOG based
classifier
McDonald et al., 2014
101,d
Different time of day: daytime,
early evening, late in the
evening (and early morning).
Accuracy and AUC for detecting drowsiness-related lane departures
Random forest steering
algorithm
Jackson et al., 201670
One night of sleep deprivation
or after a normal night of sleep
Spearman’s rank correlation: (1) standard deviation of lateral position
(2) number of crashes
(1) JDS; (2) %TEC
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
McDonald et al.,
201868,e
Driving at daytime; driving
during the late evening and
early morning.
AUC for detecting drowsy-related lane departures
MLC_DMN
Puspasari et al., 2019
59
Sleep duration was divided in
to two levels: 8 h or 4 h
Spearman’s rank correlation: (1) KSS; (2) line crossing; (3) incident
frequency
(1) Blink duration; (2)
microsleep; (3) saccadic
peak velocity
Puspasari et al., 2019
59
Sleep duration was divided in
to two levels: 8 h or 4 h
AUC for discriminating: (1) alert and low-level fatigue conditions; (2)
low-level and heavy fatigue conditions (Alert: KSS 1–5; low-level
fatigue: KSS 6–7; heavy fatigue: KSS 8–9 and experienced line crossing
at least once)
(1–2) blink duration
Wilkinson et al., 2020
81
Benzodiazepine administration
or placebo
Sensitivity and specificity for detecting driving simulator crashes
Blink duration
Aviation-related task
McKinley et al., 2011 64
Experimental testing began
after 14 h awake and continued
until 28 h of sleep deprivation
Partial correlations of reaction time to detect pop-up enemy targets
while simulated flying during target acquisition task
Approximate entropy of
pupil position
Sleep disorders
Kratzel et al., 2021 55 ,f
MWT in patients with sleep
disorders
AUC for discriminating PSG measured sleep in a MWT trial
PERCLOS
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023

Accepted Manuscript
See Table 1 for information about devices and participants; information on studies not included in Table 1 is given in footnotes. AUC: area under the
receiver operating characteristic curve; BZ: benzodiazepine; EEG: electroencephalography; EOG: electrooculography; EP: error profile; IED: inter-event
duration; JDS: Johns Drowsiness Scale; MLC_DBN: dynamic Bayesian network (DBN) that integrates multiple random forest observations from steering
angle and pedal input with maneuver-level context from vehicle speed and acceleration; MWT: maintenance of wakefulness test; %TEC: percentage of time
with eyes closed; PosAVR: positive amplitude-velocity ratio of each blink; PSG: polysomnography; PVT: psychomotor vigilance test; a Note that the AUC of
RR-interval PSD (0.02–0.08 Hz) was not significantly different from that of PERCLOS for the 25% and 75% PVT lapse increases.; b Recordings were during
looking ahead; c Device: Optalert (Sleep Diagnostics Pty Ltd., Melbourne, Australia); N = 33; Subject type: healthy adults; age range: 18–70 years; d Device: NA; N
= 72; subject type: healthy drivers; Age range: 21–68 years; e Device: FaceLab 5.0 (Seeing Machines, Canberra, Australia); N = 7 for test dataset and N = 65 for
training data; subject type: healthy adults: age range: 22–57 years (test dataset); f Device: Drowsimeter R100 (Phasya S.A., Seraing, Belgium); N = 30; Subject
type: Suspected or diagnosed narcolepsy, idiopathic hypersomnia or obstructive sleep apnea; age range: 18–80 years.
Downloaded from https://academic.oup.com/sleepadvances/advance-article/doi/10.1093/sleepadvances/zpad006/7000589 by Shaw University user on 25 January 2023