Sleepiness and safety: Where biology needs technology
Abstract
Maintaining human alertness and behavioral capability under conditions of sleep loss and circadian misalignment requires fatigue management technologies due to: (i) dynamic nonlinear modulation of performance capability by the interaction of sleep homeostatic drive and circadian regulation; (ii) large differences among people in neurobehavioral vulnerability to sleep loss; (iii) error in subjective estimates of fatigue on performance; and (iv) to inform people of the need for recovery sleep. Two promising areas of technology have emerged for managing fatigue risk in safety‐sensitive occupations. The first involves preventing fatigue by optimizing work schedules using biomathematical models of performance changes associated with sleep homeostatic and circadian dynamics. Increasingly these mathematical models account for individual differences to achieve a more accurate estimate of the timing and magnitude of fatigue effects on individuals. The second area involves technologies for detecting transient fatigue from drowsiness. The Psychomotor Vigilance Test (PVT), which has been extensively validated to be sensitive to deficits in attention from sleep loss and circadian misalignment, is an example in this category. Two shorter‐duration versions of the PVT recently have been developed for evaluating whether operators have sufficient behavioral alertness prior to or during work. Another example is online tracking the percent of slow eyelid closures (PERCLOS), which has been shown to reflect momentary fluctuations of vigilance. Technologies for predicting and detecting sleepiness/fatigue have the potential to predict and prevent operator errors and accidents in safety‐sensitive occupations, as well as physiological and mental diseases due to inadequate sleep and circadian misalignment.
Introduction
There are extensive data documenting that acute and chronic partial sleep loss, prolonged wakefulness, and waking performance at night when humans are biologically programmed to sleep, are risk factors for performance errors and accidents in a wide range of occupational settings.1-3 In addition, short sleep duration, sleep disorders and circadian misalignment have been found to associate with several physiological and mental disorders including hypertension, diabetes, obesity, depression, or cancer.4-11 Recently, two new promising technologies for managing sleepiness/fatigue risk in human systems have emerged. These include preventing fatigue by optimizing work schedules using biomathematical models of performance changes associated with sleep and circadian dynamics,12, 13 and technologies for detecting drowsy and fatigued operators on the job.14 A recent review of technologies for managing fatigue and sleepiness identified that there are significant challenges related to these and other fatigue mitigation technologies.14 There is need to establish their validity, safety value, acceptance, use adherence, and abuse potential.13-16
Fatigue is the word used throughout government, industry, labor, and the public to indicate the effects of working too long, following too little rest, and/or being unable to sustain a certain level of performance on a task.1 These issues overlap extensively with those that relate to sleepiness and its performance effects, and consequently, sleepiness and fatigue are used interchangeably in this review.
Operators' Incapacitation from Fatigue Requires Novel Solutions
Human neurobehavioral functions (e.g. alertness, attention, working memory, problem solving, reaction time, situational awareness, risk taking, etc.) are dynamically controlled by the interaction of sleep homeostatic drive and circadian regulation.17-19 When total sleep deprivation is continued for several days, the detrimental effects from sleep homeostatic drive on alertness and performance continue (nearly linear) to the increase; however, the circadian process modulates the changes daily and can mitigate some of the effects of sleep loss during times of the circadian peak.20 For example, when remaining awake for 40 h, it is a counterintuitive fact that fatigue and performance deficits are worse at 24 h than at 40 h awake. Dependence on these processes makes the prediction of neurobehavioral performance nonlinear. The nonlinearity means that performance predictions based on simple linear fatigue models, which are widely used by industry and regulatory bodies, are often grossly inaccurate. These historical limits on work time are all based on the assumption that the longer one works the more fatigued one will become. In contrast, there is extensive scientific evidence that work‐related fatigue limits should be based on the amount of sleep obtained and on circadian phase as they dynamically interact over time modulating performance capability and therefore safety. This dynamic nonlinearity in the brain's performance capability is the reason that developing and validating mathematical models that predict performance is increasingly regarded as essential. These models have increasingly assumed a critical role in fatigue risk management.12, 13
A second area of technology development concerns the development and validation of technologies for detecting fatigued operators on the job.14 There are three scientifically‐based reasons why objective sleepiness‐detection technologies are needed in safety‐sensitive operations. One reason is to inform people of when recovery sleep is essential and if possible, how much sleep is needed. There are extensive data documenting that performance deficits from sleep loss accumulate over days to high levels when daily recovery sleep is chronically inadequate.20-26 Two seminal experimental studies documented precise dose‐related effects of chronic sleep restriction on neurobehavioral performance measures in healthy adults.21, 22 In both experiments, performance deficits increased steadily across consecutive days of sleep restriction, and the less sleep chronically provided each night below 7 h, the more rapidly the performance deficits increased across days of restriction. Within 5–6 days of sleep restricted to less than 7 h, decrements in behavioral alertness increased to levels equivalent to having had no sleep at all for 24–48 h. In addition, there is scientific evidence that one night of 10 h sleep is not sufficient to recover from neurobehavioral deficits after five consecutive nights of 4 h sleep restriction.26
Another justification for technologies that detect fatigued operators stems from the fact that humans are often unable to subjectively estimate the degree of impairment of their alertness and performance due to inadequate sleep, working at night, or a sleep disorder.22, 27 A classic finding from experiments on chronic partial sleep deprivation is that people overestimated their subjective alertness and underestimated the severity of their reduced behavioral alertness and the likelihood of having performance lapses or sudden sleep onsets under conditions of chronic partial sleep loss.22 That is, people tend to believe they can overcome sleepiness either by force of will or by engaging in certain behaviors (e.g. listening to music, etc.), but these alerting stimuli have only small and short‐lived effects.28, 29 In addition, fatigue‐risk management programs that rely largely on self‐reported fatigue sleepiness are likely to miss at‐risk chronically sleep‐deprived individuals, and those at greatest risk for a performance lapse that could have serious consequences for safety (e.g. drift off of road crash).
A third reason for technologies that detect fatigued operators relates to the large and stable differences among people in the rate at which they are neurobehaviorally vulnerable to sleep loss and night work. While everyone will ultimately experience neurobehavioral deficits from sleep loss if it is sustained long enough, some individuals are highly vulnerable to performance deficits early in sleep deprivation (we labeled these Type 3 responses), while others take much longer to show deficits or manifest only moderate deficits until sleep loss is severe (labeled Type 1 responses). Still others show deficits intermediate to these two extremes (labeled Type 2 responses).30, 31 These individual differences in response to sleep loss may depend on the cognitive domain studied – an area requiring further research. Regardless of the cognitive area, they appear to be stable and trait‐like, indicative of a phenotypic response.32-34 For example, in experiments involving repeated exposure to sleep deprivation in the same subjects, the intraclass correlation (ICC) coefficient, which expresses the proportion of variance that is explained by systematic inter‐individual variability, revealed that stable responses within individuals accounted for between 58 to 68% of the overall variance in degradation of vigilant attention measured by the Psychomotor Vigilance Test (PVT: see details in fatigue detection technologies' section) between multiple sleep‐deprivation exposures.22, 31-36 Thus, healthy adults who had high lapse rates during sleep deprivation after one exposure also had high lapse rates during a second exposure (separated by weeks or months), and similarly, those with low lapse rates during one exposure had low lapse rates during a second exposure.30, 32, 34 These results strongly suggest a genetic component of different vulnerability to sleep loss.33, 37-40 A recent study by Kuna and colleagues33 of monozygotic (MZ) and dizygotic (DZ) twin pairs confirmed the genetic component of neurobehavioral vulnerability to sleep loss. They found that the ICC for PVT lapses over 38 h of sleep deprivation in MZ twin pairs was 56.2%, whereas it was 14.5% for DZ twins, showing that behavioral impairment produced by sleep deprivation is a highly heritable trait. These discoveries have resulted in a search for biomarkers that would predict the vulnerability of individuals to the neurobehavioral effects of sleep loss. Several recent studies have investigated the effects of genetic polymorphisms33, 37-39, 41-44 and neuroimaging biomarkers45-51 on inter‐individual differences in neurobehavioral vulnerability to sleep loss. It is not known if other effects of sleep loss (e.g. weight gain8) demonstrate phenotypic vulnerability. Although prediction of sleepiness is a critical goal for fatigue management, the latter also requires detection of sleepiness in real time in order to prevent imminent risk of errors and accidents. Short term fluctuation between alertness and drowsiness can occur even among individuals who are less vulnerable to sleep loss. Therefore, in addition to biomarkers for trait‐like performance vulnerability to sleep loss,52 fatigue detection technologies offer the ability to detect the immediate state of an operator.
Sleepiness/Fatigue Prediction Technologies
Mathematical models predicting sleepiness/fatigue over multiple days have received significant attention in the past two decades.12, 17, 53, 54 The two‐process model of sleep regulation17 can predict sleep timing and duration; however, this simple model failed to predict neurobehavioral effects of chronic sleep restriction.55, 56 Recent biomathematical models of neurobehavioral performance have been developed to predict behavioral alertness to both total sleep deprivation and chronic sleep restriction.54, 57 An important prediction from the model54 is that deterioration of the neurobehavioral performance converged to an asymptotically stable equilibrium when daily wake duration was below 20.2 h (3.8 h time in bed), but performance deficits increased markedly when daily wake duration was above 20.2 h (i.e. less than 3.8 h of sleep in 24 h). Another important prediction from this model is that a single night of recovery sleep is inadequate to recover from chronic sleep restriction. This prediction has been confirmed by the recent experimental findings.26
Another limitation of the previous mathematical models is that they failed to accurately predict behavioral alertness of individuals with the different phenotypic vulnerabilities to sleep loss. To address this limitation, Van Dongen et al.53 developed an adaptive Bayesian forecasting performance prediction method that uses the results of an individual's past performance to identify the values of his/her traits, and then predicts future performance, updated by a fatigue detection technology. As the number of past data points increases, the model increases the accuracy with which the trait parameters are estimated (Fig. 1). The individualized predictions more accurately predict actual future performance of each individual than does the population average prediction. The mathematical model accounting for individual differences achieves a more accurate estimate of the timing and magnitude of fatigue effects on individuals,53 which should facilitate a use of individualized countermeasures (e.g. naps, recovery sleep, caffeine intake).
Simulation using the Bayesian forecasting procedure to predict future performance of three individuals, measured with the 10‐min Psychomotor Vigilance Test (PVT), during total sleep deprivation. Performance is predicted starting from t = 44 h of wakefulness, with mean (black line) and 95% confidence intervals (vertical lines). Individual predictions are based on traits identified from prior performance measurements up to 44 h (black dots). The gray circles show the actual performance measurements during the 24 h prediction period. Figure reprinted with permission from Van Dongen and colleagues.53
Mathematical modeling is currently being used to identify work schedules that pose a sleep deprivation risk and to estimate the magnitude of the risk.13 There is recognition, however, that mathematical models developed to predict and prevent fatigue risks from sleep loss and circadian interactions have limitations. For example, they need feedback from actual values of neurobehavioral performance to improve their accuracy.53 In addition, no model can predict a momentary change of fatigue/sleepiness. Therefore, they may be only one of the important elements in a fatigue risk management system. Integrated use of sleepiness‐prediction and detection technologies holds promise as technologies that could be used to mitigate accidents and the risk of errors more effectively.
Sleepiness/Fatigue Detection Technologies
Fitness‐for‐duty tests
Vigilant attention is a requirement for a great many safety‐sensitive tasks, from operating moving conveyances, to performing many kinds of work, to detecting anomalies and threats. Reviews of cognitive performance tests have consistently found that vigilant attention tasks are among the most sensitive measures of sleep loss and circadian periodicity.58 A recent meta‐analysis investigated 70 published studies of the effects of a night of acute total sleep loss on a total of 147 cognitive tests including simple attention, complex attention, working memory, processing speed, short‐term memory and reasoning.58 This study revealed that effect sizes were largest for lapses in attention and smallest for reasoning accuracy.58 Thus, deficits in the ability to sustain attention and respond quickly are among the primary adverse effects of inadequate sleep on performance.58
Psychomotor Vigilance Test performance, in particular, has proven to be very sensitive to all types of sleep loss, while also having the advantage of virtually no learning curve or aptitude variance.59 The PVT is an example of a probed‐performance fitness‐for‐duty test.59 It is based on probing the ability of the brain to sustain attention and respond quickly (i.e. behavioral alertness) and relies on very precise measurements of repeated reaction times (RT) to a simple visual (or auditory) stimulus occurring at a predetermined inter‐stimulus‐interval (ISI) range. The PVT typically requires a button press to the onset of a visual millisecond counter. Stimuli are presented with a random inter‐stimulus interval of 2 to 10 s. The digital counter showing reaction time to the light stimulus remains visible and stops counting immediately at the subject's response. All responses are displayed digitally in milliseconds (ms), with incorrect responses (i.e. false start, incorrect key press, or keeping the button pressed) coded into the recording as errors. Basner and Dinges reported criteria for method and variables for the 10 min‐PVT.59 They also recommended that reciprocal mean 1/RT (i.e. response speed) and number of lapses should be considered as primary outcomes for the 10‐min PVT due to their superior conceptual and statistical properties and high sensitivity to sleep deprivation.59 Both the standard 10‐min PVT59 and the briefer 3‐min PVT (the Brief PVT: PVT‐B),60 which is based on a modified performance algorithm, have been extensively validated to be sensitive to both acute total and chronic partial sleep deprivation, revealing the temporal dynamics of sleep homeostatic and circadian interactions. The 10‐min PVT has become perhaps the most widely used measure of behavioral alertness owing in large part to the combination of its brevity, its high sensitivity to both acute total sleep deprivation and chronic sleep restriction, and its psychometric advantages over other neurobehavioral tests of sleepiness.59 It has also been validated as a reliable measure to identify fatigue in occupational settings61 and clinical settings62 as well as to screen for sleep apnea patients who have higher risks of fatigue‐related accidents.63, 64
However, the standard 10‐min PVT is often considered impractical for operational or clinical settings because of its duration. Neurobehavioral tests for fatigue assessment and fitness for duty not only need to be operationally and conceptually valid, reliable, sensitive, specific, generalizable, and easy to use, but also brief enough to be acceptable for the target population and to allow for repeated administration in operational environments. To meet these criteria, two shorter‐duration versions of the PVT (with modified algorithms for performance evaluation) have been developed with extensive validation for their sensitivity to both acute total and chronic partial sleep deprivation. These are the Brief PVT (PVT‐B)60 and the Adaptive‐Duration Version of the PVT (PVT‐A).65
The PVT‐B has ISIs decreased from the standard 2–10 s of the 10‐min PVT, to 1–4 s and reduced lapse threshold from 500 to 355 ms.60 The PVT‐B has been shown to track the standard 10‐min PVT closely over time in experiments on both total sleep deprivation and chronic sleep restriction.60 PVT‐B test duration was decreased 70% relative to the 10‐min PVT; its effect size for sensitivity to sleep loss was decreased by only 22.7%.60 This is an acceptable trade‐off between task duration and sensitivity. In a laboratory study of work performance to determine if the PVT‐B had potential as a fitness‐for‐duty test, it was demonstrated that PVT‐B closely tracked fatigue‐related threat‐detection performance decrements on a simulated luggage‐screening task.66 Performance on the PVT‐B and the simulated luggage‐screening task covaried over a 34 h period of total sleep deprivation. This is a particularly important finding because the threat‐detection task has high fidelity to what operators must do while screening luggage through X‐ray machines. Thus, the PVT‐B has the potential to predict operationally‐relevant performance relative to vigilance work.
Although PVT‐B may be a useful tool for assessing behavioral alertness in settings where the duration of the 10‐min PVT is considered impractical, the shorter PVT versions seem to be too short to detect relevant deterioration in vigilant attention in subjects with moderate impairment whose performances deteriorate only later during the test, whereas the longer versions may be unnecessarily long for other subjects who are apparently fully alert or severely impaired.65 The adaptive PVT (i.e. PVT‐A) is a modified PVT with a duration dependent on the subject's performance.65 Thus, in contrast to the fixed durations of the PVT and PVT‐B, the PVT‐A65 duration is variable. It stops sampling once it has gathered enough information to correctly classify performance as high, medium, or low, according to the number of lapses and false starts. In a validation experiment, test duration of the PVT‐A averaged less than 6.5 min (SD 2.4) for a training dataset and 6.4 min (SD 1.7) for a validation dataset. In addition, the PVT‐A was shown to be highly accurate, sensitive, and specific relative to 10‐min PVT performance. Thus, the adaptive‐duration strategy of the PVT‐A may be superior to a simple reduction of PVT duration. Future studies are needed to show its feasibility and usefulness in professional screeners and operational environments as a fitness‐for‐duty test.
Online operator monitoring
Fitness‐for duty tests hold the promise of detecting the state of sleep‐related fatigue in populations at risk for accidents and errors due to fatigue‐inducing work schedules. However, as noted above, the neurobehavioral effects of sleep loss and circadian periodicity follow a non‐linear time course within and between days, as well as more transient evoked effects on alertness from body posture, social interaction, caffeine, etc. Therefore, using biomathematical models augmented with online operator monitoring may be a more comprehensive way to detect fatigue relative to work. The following is an example of one type of continuous monitoring of operator fatigue based in sleepiness, using a measure of slow eyelid closures (i.e. slow blinks) referred to as PERCLOS (proportion of time that the eyes are closed over a certain interval).16, 67-72 This example illustrates the criticality of the validation science that must be undertaken as an initial first step toward developing a truly reliable unobtrusive measure of sleepiness.
To validate PERCLOS and several other approaches, Dinges and colleagues16, 67 systematically evaluated the validity of a number of putative sleepiness‐detection technologies. These included brain wave (electroencephalogram [EEG]) algorithms, eye‐blink rate devices, a measure of slow eyelid closures (i.e. PERCLOS), and a head position sensor, as well as individuals' own ratings of their sleepiness. In a series of tightly controlled, double‐blind experiments, they evaluated the extent to which each technology detected the alertness of subjects over a 40‐h period of wakefulness, as measured by PVT lapses of attention – a well‐validated measure of behavioral alertness.16, 67 Each putative fatigue‐detection technology was time‐locked to PVT performance in a manner that permitted precise determination of whether a given technology could reliably track minute‐by‐minute (across a normal day and during nocturnal and diurnal periods of sleep deprivation), the waxing and waning of alertness as evident in PVT lapses of attention. The evaluation for the minute‐by‐minute recordings from each technology was done by the respective developer of each technology, blind to PVT performance (and the latter was scored for performance lapses blind to each technology score of alertness for each minute). A biostatistician then fit the prediction of alertness from each technology to the PVT performance data for each subject across a 42‐h period of evaluation. This resulted in a measure of statistical coherence for each technology for each study participant evaluated. Human‐scored PERCLOS proved superior to all other detection technologies in blindly predicting when PVT lapses of attention were occurring across the 42 h awake time each subject underwent. The initial results reported in Dinges et al.16 were subsequently replicated for a retinal reflectance measure of PERCLOS.67 As shown in Figure 2, only PERCLOS reliably and accurately tracked PVT lapses of attention in all subjects, outperforming not only all the other technologies, but also subjects' own ratings of their fatigue and alertness in both validation trials.16, 67
Mean percent time of slow eyelid closures (PERCLOS) coherence for Psychomotor Vigilance Test (PVT) lapse frequency across 42 h of waking (triangles), as a function of the time base used to define an epoch. A distance‐weighted least squares function was fit to the data. PERCLOS was measured by a human scoring videos of slow eyelid closures (Experiment 1) and by infrared retinal reflectance (Experiment 2, CMRL). In both experiments it had much higher coherence with PVT performance lapses of attention (i.e. high sensitivity to behavioral alertness) than any other technology evaluated in the experiments (i.e. two different electroencephalogram [EEG] algorithms [EEG‐1, EEG‐2), two different eye blink technologies [Eye blink‐1, Eye blink‐2], and head movement sensor technology [Head sensor]). PERCLOS was also a better predictor of alertness than subjects' self‐reports of sleepiness by a visual analogue scale (i.e. VAS sleepiness). The accuracy of PERCLOS predictions of PVT performance increased as the time base for integrated assessments increased from 1 to 20 min. More recent work also supports the accuracy of PERCLOS for unobtrusive detection of sleepiness while performing a behavioral maintenance of wakefulness test70 and the PVT.71 Figure reprinted from Dinges and colleagues.16, 67
More recent studies have compared accuracies for predicting vigilance deterioration among several measurements including EEG frequency band activities, heart rate variability, and ocular variables (saccade, slow eye movement, pupil, blink, or eyelid closure).70, 71 The experiments also found that PERCLOS was the most effective indicator of sleepiness‐based fatigue among the variables evaluated.70, 71 Dinges et al.73-76 are now developing a new technique that involves precise and completely unobtrusive tracking of PERCLOS in real time, using optical computer recognition.
Another example of online operator monitoring technology is the Johns Drowsiness Scale (JDS; scores ranging from 0 to 10, where 0 = very alert and 10 = very drowsy) based on a weighted combination of several sleepiness indicators derived from ocular measures such as blink duration and amplitude‐velocity ratios during the closing and reopening phase of blinks measured by infrared reflectance oculography.77, 78 The JDS score was shown to track performance levels during vigilance attention tasks, and a driving simulator task as well as alertness levels after caffeine ingestion.77-82 In addition, higher JDS scores (≥4.5) were associated with self‐reported inattention during on‐road driving events in nurses commuting to and from night and rotating shifts.83 However, the JDS requires wearing special glasses, which may be a deterrent to its use in certain settings.
Two independent studies using 40 h of continuous wakefulness under constant routine have investigated the accuracies (area under the receiver‐operating characteristic curve: AUC ranging from 0.5 to 1.0; higher value is better) of PERCLOS or JDS to identify a threshold increase (>25%, >50%, and >75%) in the number of PVT lapses, measured relative to each subject's performance during baseline (first 16 h of wakefulness).71, 80 The results showed that AUC for the PERCLOS and JDS are 0.89–0.91 and 0.74–0.76, respectively.71, 80 Although the procedures for measuring the vigilance were not identical (auditory or visual PVT, 1 h or 2 h intervals of test bout, etc.), these results indicate that the accuracy of PERCLOS was higher than that of JDS. Future studies will be needed to compare their accuracies in the same protocol.
One study showed that PVT lapses occur when eyes are open,84 meaning deteriorated vigilance can occur even during no sign of PERCLOS.70 Missing detection of deteriorated vigilance (false negatives) potentially causes accidents, and inappropriate warning of decreased vigilance in alert persons (false positives) may decrease compliance for use of the technologies. There need to be continued improvements of the accuracy of online operator fatigue monitoring.
Field study of fatigue‐detection technologies
In one of the relatively few fatigue‐monitoring studies in over‐the‐road commercial truck drivers, Dinges and colleagues investigated whether feedback from fatigue‐detection technologies would help truck drivers maintain their alertness in actual driving conditions.85 The technologies included driving performance variables (e.g. lane tracking variability), PERCLOS, head sensor, wrist actiwatch and the 10‐min PVT test. The results from this study revealed that the drivers felt the fatigue detection devices informed them of their fatigue levels and prompted them to acquire more sleep on their days off duty. In fact, the wrist actigraphy data confirmed that when receiving feedback on their alertness levels, drivers increased their sleep by an average of 45 min on days off duty.85 This is a remarkable and unexpected outcome, and it suggests another purpose for fatigue detection technologies in the workplace – namely to urge operators to obtain longer recovery sleep. If we could use fatigue management technologies to warn drivers when they are getting sleepy and to encourage them to get off the roadway, it may be possible to reduce the risk of sleepiness‐related accidents and errors.
Another example of fatigue detection technologies used in real‐world operational environments is vigilance monitoring of astronauts who stay long term in the International Space Station (ISS). Spaceflight Cognitive Assessment Tool for Windows (WinSCAT) has been used to evaluate neurobehavioral performance levels of astronauts in the ISS;86 however, the WinSCAT requires about 30 min of crew time. Therefore, the test is not suitable to evaluate astronauts' neurobehavioral performance within a day or every day. The PVT‐B also has been studied on ISS to evaluate astronauts' vigilance levels.87 Importantly, a comprehensive yet brief performance test (now referred to as COGNITION) has been developed to evaluate several neurocognitive and emotional domains of astronauts' performance on board the ISS.88 These approaches offer a way to quickly and reliably assess not only behavioral alertness, but also a range of cognitive functions that may be affected by sleep loss.
Conclusion
Technology for predicting and evaluating whether operators have sufficient performance capability (relative to sleep need and circadian timing) prior to beginning or during their work is important to prevent accidents and errors due to the presence of fatigue‐related neurobehavioral deficits. In addition, complementary use of these technologies including mathematical models, fitness‐for duty tests, or online operator monitoring might be more effective to find the risk of accidents and errors quickly and accurately. Such technologies allow operators to use countermeasures to mitigate sleepiness and fatigue before starting or continuing their work, contributing to reduce operational errors and accidents due to neurobehavioral deficits from sleep loss and circadian misalignment. Furthermore, considering that sleep loss is a risk factor for several physiological and mental disorders,4-11 the urging effect of fatigue‐management technologies to inform people of the need for recovery sleep might also contribute to prevent developing various diseases associated with inadequate sleep.
Disclosure
This was not an industry supported study. Dr Abe and Dr Basner have no financial conflicts of interest. Dr Mollicone is president and CEO of Pulsar Informatics. Dr Dinges is compensated by the Associated Professional Sleep Societies, LLC, for serving as Editor in Chief of SLEEP and has received compensation for serving on a scientific advisory council for Mars, Inc.
Acknowledgments
The time and effort required to write the review were supported by JSPS Postdoctoral Fellowships for Research Abroad and KAKENHI Grant Number 22730598 (T. Abe); NIH grant R01 NR004281 (D.F. Dinges); National Space Biomedical Research Institute through NASA NCC 9‐58 (D.F. Dinges); and by the Office of Naval Research and the Navy BUMED Advanced Medical Development Program through contracts N65236‐09‐D‐3809, N00014‐10‐C‐0392, N00014‐11‐C‐0592, and N62645‐12‐C‐4004 (D. Mollicone).
