Administration of blue light in the morning and no blue-ray light in the evening improves the circadian functions of non-24-hour shift workers

ABSTRACT In modern 24-hour society, various round-the-clock services have entailed shift work, resulting in non-24-hour schedules. However, the extent of behavioral and physiological alterations by non-24-hour schedules remains unclear, and particularly, effective interventions to restore the circadian functions of non-24-hour shift workers are rarely explored. In this study, we investigate the effects of a simulated non-24-hour military shift work schedule on daily rhythms and sleep, and establish an intervention measure to restore the circadian functions of non-24-hour shift workers. The three stages of experiments were conducted. The stage-one experiment was to establish a comprehensive evaluation index of the circadian rhythms and sleep for all 60 participants by analyzing wristwatch-recorded physiological parameters and sleep. The stage-two experiment evaluated the effects of an intervention strategy on physiological rhythms and sleep. The stage-three experiment was to examine the participants’ physiological and behavioral disturbances under the simulated non-24-hour military shift work schedule and their improvements by the optimal lighting apparatus. We found that wristwatch-recorded physiological parameters display robust rhythmicity, and the phases of systolic blood pressures and heart rates can be used as reliable estimators for the human body time. The simulated non-24-hour military shift work schedule significantly disrupts the daily rhythms of oxygen saturation levels, blood pressures, heart rates, and reduces sleep quality. Administration of blue light in the morning and no blue-ray light in the evening improves the amplitude and synchronization of daily rhythms of the non-24-hour participants. These findings demonstrate the harmful consequences of the non-24-hour shift work schedule and provide a non-invasive strategy to improve the well-being and work efficiency of the non-24-hour shift population.


Introduction
Almost all organisms, including humans, possess an internal timekeeping mechanism known as the circadian clock that dictates the timing of various body functions (Patke et al. 2020;Roenneberg et al. 2022;Wang 2018).The circadian clock contributes to the regulation of locomotor activity, body temperature, hormone secretion, metabolism, brain functions, and the sleep-wake cycle (Bass and Takahashi 2010; Cassone et al. 1993;Karatsoreos et al. 2011;Klerman et al. 2002;Zhong et al. 2023).While the master clock is situated in the suprachiasmatic nuclei (SCN) in the hypothalamus of the brain, the peripheral clocks also operate in almost all the organs, like the heart, the pancreas, and the liver (Cagampang and Bruce 2012;Storch et al. 2002).These circadian oscillators act synchronously to generate rhythmic functions aligned with the 24-hour cycle of the day.The photosensitive and melanopsin-expressing retinal ganglion cells send light information to the SCN (Gooley et al. 2001).These specialized retinal cells are maximally sensitive to the blue spectrum (464 nm) of light and are essential for entraining the circadian rhythms to the environmental light-dark cycle (Brown et al. 2022).The SCN projects to the pineal gland, which rhythmically releases melatonin, a sleep-promoting hormone (Cassone et al. 1993).
In modern 24-hour society, round-the-clock services have incorporated shift work into daily work routines (Boivin and Boudreau 2014).Under some special circumstances, for instance, non-24-hour rotating watchkeeping systems have strictly been enforced on military operational units such as submarines (Beare et al. 1981;Colquhoun et al. 1979;Duplessis et al. 2007;Kelly et al. 1999).Shift work and light at night (LAN) have long been known to affect human health, and previous studies have demonstrated that shift work or non-24-hour work regimes may trigger sleep disorders (Boivin and Boudreau 2014;Condon et al. 1988;Short et al. 2016), cardiovascular diseases (Chellappa et al. 2019), obesity (Huang et al. 2021b), diabetes mellitus (Ikegami et al. 2019), mental disorders and cognitive deficits (Chellappa 2020;Condon et al. 1988;Fischer et al. 2019), and cancers (Brudnowska and Peplonska 2011).As such, it is important to examine the circadian functions and sleep quality of the shift work population.The traditional gold standard for evaluating the human body clock is the DLMO (Dim light melatonin onset) method (Klerman et al. 2002).However, this method requires collecting a large quantity of blood or saliva samples for laboratory tests.The core body temperature (CBT) of humans also displays robust rhythmicity, but it entails implanting a body temperature detection sensor or taking pills (Mendt et al. 2021).Therefore, there is an urgent need to develop a relatively reliable and straightforward method for evaluating the rhythm of the shift workers (Dijk and Duffy 2020; Wittenbrink et al. 2018;Wu et al. 2018Wu et al. , 2020)).In this aspect, non-invasive and sophisticated smart wristwatches have been developed and increasingly employed to record numerous circadian and sleep features of humans (Liang et al. 2018;Xie et al. 2018;Zhang et al. 2022).
Light at night (LAN) has been known to act through the eye-SCN-pineal pathway to suppress and/or shift the daily melatonin rhythms (Giudice et al. 2018;Hester et al. 2021;Hunter and Figueiro 2017).In addition, light photoreception and its effects on the circadian system are known to be sensitive to light characteristics, including both wavelengths and intensities (Sunde et al. 2020).For instance, blue-enriched (464 nm) light exposure in the evening/night not only exerts more substantial inhibitive effects on melatonin release than standard (550 nm) light (Sasseville et al. 2006) but also reduces the slow-wave activity (SWA) during the NREM (non-rapid eye movement)-REM (rapid eye movement) sleep cycle (Munch et al. 2006), probably due to the reduced melatonin levels.
Even though being intensively investigated in the past decades, the details of physiological and behavioral effects of circadian misalignment by shift work or non-24-hour work regimes remain unclear (Huang et al. 2021a;Ma et al. 2019), and particularly, interventions to improve the circadian functions of non-24-hour shift workers are rarely explored.Critical questions are how the light spectrum affects the circadian rhythms of non-24-hour shift workers and what are possible strategies for intervening circadian misalignment and alleviating its harmful effects.Here we investigated the impact of a non-24-hour military shift work schedule on daily rhythms of oxygen saturation levels, blood pressures, heart rates, wrist temperature, locomotor activity, and melatonin secretion, as well as sleep.We further established an intervention strategy for alleviating the effects of the non-24-hour regime.

Subjects
For this study, 60 male participants were recruited using an online messenger platform (WeChat) at Soochow University, Suzhou, China.These selected participants, aged 22-36 years, had no reported history of psychiatric, neurological, and sleep disorders, with normal optesthesia (Supplementary Table S1).Most of the participants were graduate students who majored in sciences and were required to conduct scientific experiments.Participants were not engaged in any prior shift work.Participants all lived in student dormitories at Soochow University during the experimental period.During the experiment, except otherwise indicated, the participants were under a common LED light source with a wavelength of 380-780 nm at 200-400 lux in their dormitories before going to bed at night.All research protocols were performed in accordance with the guidelines approved by the Ethics Committee of Soochow University (#SUDA20200730H03).All the participants signed written informed consent.

Experimental design and procedures
The study was conducted in three stages in five weeks, from August to September 2020.Participants combined workdays and off-duty days during the three stages of the experiment, and their schedules were not restricted to only normal social interaction and work duty.The participants were advised to wake up between 07:00 and 08:00 and go to bed between 23:00 and 24:00 in stageone and stage-two experiments.The stage-one experiment was to establish a comprehensive evaluation index of circadian rhythms and sleep for all 60 participants.We recorded the oxygen saturation levels, blood pressures, heart rates, wrist temperature, steps, and sleep with wristwatches (Huawei Honor i5 and Pulibang e66, Shenzhen, China) for ten consecutive days under the normal 24-hour living condition.After the completion and evaluation of the first stage, 30 participants with relatively poor sleep quality were selected for the stage-two experiment, while 16 participants with relatively good sleep quality were selected for the stagethree experiment (Figure 1a).
For the stage-two experiment, all 30 participants were required to wear either blue (480 nm, 12 lux, Tc: 49999 K, 0.10 W/m 2 ) or green (510 nm, 30 lux, Tc: 9330 K, 0.09 W/m 2 ) light smart eyeglasses (PEGASI, Shenzhen, China) for 30 minutes after waking up in the morning and were exposed to different intensities (500, 200 or 60 lux) of the high spectrum (630 nm; no blue-ray, Tc: 1800 K, 0.11-0.136W/m 2 ) light (MIDIAN, Jiaxing, China) or normal white light (60 lux, Tc: 5000 K, 0.38 W/m 2 , MIDIAN, Jiaxing, China) for 30 minutes in the evening before sleep.The light intensity and wavelength were measured vertically at the position of the participant's eyes by an illuminometer Hongpu,Hangzhou,China).The smart glasses do not have lenses, and myopic participants can wear both their own myopia glasses and smart glasses simultaneously.The light, harmless to the eye, radiates upward from the bridge of the nose.The light source of smart glasses is approximately three cm from the retina.The non-blue light source is approximately 0.5 m away from the eye.Subsequently, we had six groups, each with 5 participants: Group 1. Blue eyeglasses in the morning and 500 lux no blue-ray light in the evening, Group 2. Blue eyeglasses in the morning and 200 lux no blue-ray light in the evening, Group 3. Blue eyeglasses in the morning and 60 lux no blue-ray light in the evening, Group 4. Blue eyeglasses in the morning and 60 lux white light in the evening, Group 5. Green eyeglasses in the morning and 60 lux no blue-ray light in the evening, and Group 6. Green eyeglasses in the morning and 60 lux white light in the evening.We also recorded oxygen saturation levels, blood pressures, heart rates, wrist temperature, step counts, and sleep of all 30 participants via wristwatches for ten consecutive days under the normal 24-hour living condition, the same condition as one in the stage-one experiment (Figure 1b,c).The ENLIGHT Checklist form was filled out to show the characteristics of our study and the light used (Supplementary Table S2), as recommended (Spitschan et al. 2023).
The stage-three experiment was to examine the physiological and behavioral disturbances of the participants under the simulated non-24-hour shift work schedule of a military operational unit for six consecutive days, as well as the effects of the optimal lighting apparatus established from the stage-two experiment on them.These 16 participants were randomly divided into two groups, each with 8 participants.One group was treated with the optimal lighting apparatuses, i.e. participants were treated with a no-blue light source for half an hour before each rest and wore blue light glasses half an hour after waking.The other group was without such light treatments as a control.The non-24-hour shift schedule is as follows: on the first day, work from 00:00 h to 03:00 h, leisure and sleep from 03:00 h to 12:00 h, work from 12:00 h to 15:00 h, relaxation and sleep from 15:00 h to 21:00 h, and work from 21:00 h to 24:00 h; on the second day, leisure and sleep from 00:00 h to 07:30 h, work from 07:30 h to 12:00 h, relaxation and sleep from 12:00 h to 18:00 h, work from 18:00 h to 21:00 h, leisure and sleep from 21:00 h to 03:00 h on the third day; work from 03:00 h to 07:30 h on the third day, relaxation and sleep from 07:30 h to 15:00 h, work from 15:00 h to 18:00 h, leisure and sleep 18:00 h to 00:00 h on the fourth day; and repeat the three-days cycle (Figure 1b,d,e).The participants completed the assigned smartphone and computer operation tasks and other normal routines under a common LED light source with a wavelength of 380-780 nm at 200-400 lux in their dormitories during working time, while they ate food, relaxed, or slept during rest time.Similarly, we also recorded oxygen saturation levels, blood pressures, heart rates, wrist temperature, step counts, and sleep of all 16 participants via wristwatches for consecutive six days (Figure 1d,e).

Recording oxygen saturation levels, blood pressure, heart rates, wrist temperature step counts, and sleep of participants with wristwatches
The oxygen saturation, blood pressure, heart rate, wrist temperature, and step parameters were recorded with the wristwatch (Pulibang e66, Shenzhen, China), the automatic measurement was set every half hour, and data was obtained through SmartHealth APP (Pulibang, Shenzhen, China).Excluding the conditions when participants played basketball, swam, bathed, or charged the battery for the wristwatch, wearing it for more than 22 hours a day was regarded as valid data.Huawei wristwatch (Honor i5) was used to record sleep parameters, including the total length of sleep (including daytime sleep and nap), sleep latency, number of wake periods, number of sleep transitions, the average length of wakefulness bout, and other parameters.Sleep latency refers to the length of time from the beginning of falling asleep to the end of the first light sleep.The average length of each round of deep sleep is the total deep sleep time divided by the deep sleep bout number.While the wristwatch also records sleep data during the day, only sleep for more than three consecutive hours is considered scientific sleep monitoring, as defined by the Huawei Sports Health APP.Specifically, multiple parameters, including REM sleep, can be obtained for scientific sleep monitoring, while only sleep duration can be obtained for non-scientific sleep monitoring.Sleeping less than 3 hours is regarded as a nap.These two types of wristwatches were worn for at least 22 hours each day during the experimental period.

Blood sample collection and separation of human peripheral blood lymphocytes
Negative pressure anticoagulation tubes with sodium citrate and a light blue cap were used for blood collection.The blood samples were collected between 9:30 h-10:30 h and transferred into 15 ml sterile centrifuge tubes containing 3 ml of the lymphocyte separation solution (TBD, Tianjin, China) for centrifuging at 20°C at 400 g (RCF) for 10 minutes.After centrifugation, the second white ring layer from top to bottom was pipetted into a new 15 ml sterile centrifuge tube.Then 10 ml of the cleaning solution was added, and the solution was centrifuged at 250 g for 10 minutes.Then, the supernatant was discarded, and the visible white cell precipitate was kept.500 µl of the washing solution was added to resuspend the cells, which were pipetted into a 1.5 ml RNA-free EP tube, and centrifuged at 300 g for 10 minutes.The supernatant was discarded, the white precipitate was kept, 1 ml of TRIzol (Invitrogen, Carlsbad, CA, USA) was added, and the peripheral blood lymphocytes were stored in a freezer at −80°C.

Extraction and purification of RNAs
Total RNAs were extracted with TRIzol from these peripheral blood lymphocytes and purified with RNA clean beads (Vazyme, N412, Nanjing, China).RNA integrity was examined on agarose gel, and RNA concentrations were determined with a Nanodrop 2000 spectrophotometer (Thermo Scientific, Waltham, MA, USA), and RNAs were stored in a freezer at −80°C.

NanoString-based determination of the body time
For NanoString-based assays, we selected 12 candidate time-telling genes, including PER1, PER2, PER3, CRY1, NR1D1, NR1D2, CRISPLD2, FKBP4, LGALS3, ELMO2, KLF9, HSPH1 and four housekeeping genes GAPDH, HPRT1, PPIA, PMSB2 as controls (Wittenbrink et al. 2018).The custom-designed probes included a 3′-end biotinylated capture probe and a 5′-fluorescencebarcoded reporter probe for each gene (Supplementary Table S5).The quality of the RNA samples was examined again with Agilent TapeStation 4200 Electrophoresis (Agilent Technologies, Santa Clara, CA, USA).Hybridization of the probes with 250 ng monocyte RNA was carried out according to the manufacturer's instructions (NanoString Technologies, Seattle, WA, USA).Raw expression data were obtained using a NanoString nCounter Digital Analyzer (NanoString Technologies, Seattle, WA, USA).Normalization was carried out in 3 steps according to the Bioconductor package NanoStringQCPro: (a) normalization by the arithmetic mean of the positive spikein controls, (b) subtraction of the mean of the negative controls, and (c) normalization by the geometric mean of the four housekeeping genes.NanoString experiments were conducted with the assistance of Precision Scientific Biomedicine (Suzhou) Co., Ltd.(Precision Scientific, Suzhou, China).

Statistical analysis
Rhythmicity analysis of all the wristwatch-recorded parameters was performed on the Biodare2 website (https://biodare2.ed.ac.uk/) (Zielinski et al. 2014).JTK (Jonckheere-Terpstra-Kendall)-Cycle was used with the preset COS_2H waveforms.For the period analysis, the input data was no dtr with the expected periods from 20 h to 28 h and the MFourFit method.Pearson correlation analysis of these parameters was conducted with OriginPro2022b.*p < 0.05, **p < 0.01, ***p < 0.001.

Participant information and their PSQI scores
For the stage-one experiment, we selected 60 male participants, with an average age of 24.63 ± 2.42 years, the oldest was 36 years old, and the youngest was 21 years old; the height was between 160 cm and 188 cm; and the weight was 48-90 kg (Supplementary Table S1).The Pittsburgh Sleep Quality Index (PSQI) scores of these 60 participants showed that one had the lowest score of 0 points and one had the highest score of 12 points, with an average of 4.833 ± 2.531 (Supplementary Table S1).

Robust rhythmicity of wristwatch-recorded physiological parameters of the participants
We recorded oxygen saturation (OS) levels, systolic blood pressures (SBP), diastolic blood pressures (DBP), heart rates (HR), wrist temperature (WT), and step counts (Steps) of all these 60 participants in stageone experiment via wristwatches for ten consecutive days.We first analyzed which parameters are rhythmic and how consistent these rhythmic parameters are.
Rhythmic analysis of oxygen saturation (OS) levels showed that most participants displayed a period of around 24 hours, peaking between 12:00 h and 18:00 h, except for 5 participants (Figure 2a-c).Further, systolic blood pressures (SBP) and diastolic blood pressures (DBP) display robust circadian rhythmicity with a nearly 24-h period, peaking around 12:00 h and 18:00 h, except for a few participants (Figure 2d-i), while heart rates (HR) of almost all the participants display robust circadian rhythmicity with a nearly 24-h period, also peaking around 12:00 h and 18:00 h, except for only three participants (Figure 2j-l).However, step counts (Steps) and wrist temperature (WT) don't display robust rhythmicity, and their phases are not consistent (Figure 2m-r).The different lifestyles, for instance, some preferring to do more outdoor activities than others among the participants, likely resulted in the notable discrepancies in the phases of wrist temperature and steps.

Wristwatch-recorded systolic blood pressures and heart rates serve as reliable estimators of the human body time
To examine the reliability of wristwatch-recorded physiological parameters, we selected the six physiological parameters of two participants, P3 and P4, in the stageone experiment for detailed analyses (Figure 3a).We conducted Pearson correlation analysis of oxygen saturation levels, systolic blood pressures, diastolic blood pressures, heart rates, wrist temperature, and steps of the P3 and P4 recorded in ten consecutive days in the stage-one experiment and found that the correlation between systolic blood pressures and diastolic blood pressures was the highest (r = 0.96, p < 0.001) (Figure 3b).Because wrist temperature is markedly affected by the external environment, it has no significant correlation with the other parameters (p > 0.05) (Figure 3b).
The multi-day wristwatch-recorded data allow for calculating human circadian clocks' period, phase, and amplitude.We then analyzed the phases of the six physiological parameters of these 60 participants and found that the phases of oxygen saturation levels, systolic blood pressures, diastolic blood pressures, and heart rates are highly consistent among these 60 participants, while wrist temperature and the number of steps show relatively large phase differences (Figure 3c).
We also compared the consistency of period, phase, and amplitude of these six wristwatch-recorded parameters of the 60 participants (Figure 3d) and found that the period consistency of oxygen saturation levels, systolic blood pressures, diastolic blood pressures, heart rates, and steps of most participants is very high, all with nearly 24 hours, except for four people (6.67%).The phase consistency of these five physiological parameters of most of the participants was also very high, except for three people (5%).The amplitudes of oxygen saturation levels, systolic blood pressures, diastolic blood pressures, and heart rates of all the participants maintain highly similar consistency, and particularly, the amplitude of heart rates is higher than those of the other three parameters (Figure 3d).Notably, the number of steps and wrist temperature lack consistency in the period, phase, and amplitude among these 60 participants (Figure 3d).Hence, while wrist temperature and the number of steps display masking effects, the other multi-day physiological parameters recorded by wristwatch have little masking effects.
We also observed that the phases of systolic blood pressures and heart rates of these 60 participants exhibit the highest consistency (Figure 3d).Therefore, we posit that these two physiological parameters can be used as reliable estimators of human circadian rhythms.

Administration of blue light in the morning and no blue-ray light in the evening improves the circadian amplitude and sleep quality
For the stage-two experiment, 30 participants were randomly divided into six groups, and 5 participants in each group were treated with different combinations of 30-minute blue or green light in the morning and 30minute no blue-ray light or ordinary incandescent light in the evening for ten consecutive days, respectively, as described above (Figure 1a).The peak spectrums of the blue glasses (Figure 4a, top) and the green glasses (Figure 4a, middle) are around 480 nm and 510 nm, respectively.In comparison, the peak spectrum of the sleep-promoting lamps is approximately 630 nm beyond the blue range (Figure 4a, bottom).
We compared the phases of both systolic blood pressures and heart rates of the participants from six groups and found that Group 3 and Group 5 displayed more consistent phases of both systolic blood pressures and heart rates than the other four groups (CV = 4.78% in Group 3 and CV = 2.98% in Group 5) (Figure 4b, Supplementary Figure S1).In addition to period and phase, the amplitude is also an important factor reflecting the stability of the circadian clock.Group 3 had a significantly enhanced amplitude ratio (p < 0.05, Figure 4c), confirming that Group 3 is the best lighting condition.
We also compared wristwatch-recorded sleep characteristics of the same participants in the stage-one and the stage-two experiment.Intriguingly, while the total sleep time of the same participants in the stage-one and the stage-two experiments in each group didn't differ, the average sleep time increment of the participants in Group 3 appeared longer than theirs in the stage-one experiment (p = 0.07, Figure 4d).The five participants in Group 3 with shortened pre-sleep latency (Figure 4e).Further, the number of sleep and wake transitions of the participants in Group 3 appeared to be reduced compared to theirs in the stage-one experiment (Supplementary Table S3), indicating that they had good and stable sleep.Together, we selected the lighting conditions in Group 3, namely, wearing blue light glasses in the morning with the lowest exposure to no blue-ray light in the evening, as the best lighting apparatus for the stage-three experiment.

Reduced sleep quality and disrupted rhythmicity of physiological parameters of the participants under the non-24-hour military shift work schedule
The selected 16 participants carried out the non-24hour military shift work schedule (Figure 1).Intriguingly, the SWL group had a longer average sleep duration tendency than the SW group (p = 0.09, Figure 6a).Even though the same participants had dramatically reduced their sleep duration in the stage-three experiment compared with the stage-one experiment, the overall sleep quality of the SWL showed a better tendency than that in the SW group (p = 0.09, Figure 6b), indicating that the lighting apparatus played a role in improving sleep deficit caused by the non-24hour shift work (Supplementary Table S4).
We also compared each participant's multi-day wristwatch-recorded physiological parameters between pre-shift (stage-one experiment) and post-shift (stagethree experiment).Even though both systolic blood pressures (Figure 5a,b) and heart rates (Supplementary Figure S2a,b) still maintained rhythmicity after the stage-three experiment, more participants whose phases were significantly shifted away from the pre-shift phase in the SW group than in the SWL group (Figure 5a,b  and Supplementary Figure S2a,b).The amplitudes of systolic blood pressures and heart rates of the participants of the SW group were lower than those in the SWL group, indicating that the lighting apparatus can partially rescue the alterations of the circadian clock caused by the non-24-hour shift schedule (Figure 5c and Supplementary Figure S2c).However, oxygen saturation levels didn't display such differences between the SW and SWL groups (Supplementary Figure S3).Further, the phases of systolic blood pressures, diastolic blood pressures, heart rates, and oxygen saturation levels of the participants who carried out the stagethree experiment maintained many statistically significant positive correlations (8 of the 16 comparisons, r > 0.5, p < 0.05) in the stage-one experiment.In contrast, only DBP in SW and DBP in SWL of the same participants showed a statistically significant positive correlation (r > 0.5, p < 0.05) in the stage-three experiment, suggesting that six days of the non-24-hour shifts suffice to desynchronize these physiological parameters of some participants (Figure 5d,e).In addition, the phases of these four physiological parameters of the 8 SWL group participants displayed one modest positive correlation between OS-1 and SBP-3 (r = 0.65, p < 0.05) and three low positive correlations (OS-1 vs. HR-3, HR-1 vs. DBP-3, and DBP-1 vs. OS-3) (r ≥ 0.38, p < 0.05) (Figure 5f,g).In contrast, the phases of these four physiological parameters of the 8 SW group participants displayed only one low positive correlation between SBP-1 and OS-3 correlation (r = 0.33, p < 0.05), implicating that the lighting apparatus was able to resynchronize the decoupled physiological parameters caused by the non-24-hour shift work.Therefore, these results indicate that the lighting apparatus can promote the stability and synchronization of the circadian rhythms of the main physiological parameters.
To accurately compare the changes of each parameter before and after the non-24-hour shift, we analyzed the changes in oxygen saturation levels, systolic blood pressures, diastolic blood pressures, and heart rates of all participants in the stage-three experiment (Supplementary Figure S4) and showed most of the participants still maintained the circadian rhythm after the non-24-hour shift work, but their amplitudes or periods were altered (Supplementary Figure S4).Together, these results indicate that six days of the non-24-hour military shifts suffice significantly to reduce sleep quality and disrupt the rhythmicity of key physiological parameters of the participants, and the light apparatus can improve the circadian functions by stabilizing and resynchronizing the circadian rhythms of the main physiological parameters.

Determining the human body time of the participants by the NanoString-based method
As shown above, wristwatch-recorded systolic blood pressures and heart rates can reliably estimate the human body time (Figures 2 and 3).A NanoStringbased method was developed to estimate the intrinsic body time by analyzing the expression levels of multiple genes in a single blood sample taken at any time of the day (Wittenbrink et al. 2018).To examine the effectiveness of interventions in the non-24-hour shift work, we also conducted the NanoString-based analysis of participants' single blood samples and showed that the participants' body times were relatively consistent before the shift work but delayed more than 1 hour after the shift work compared with those before the shift work.Intriguingly, the eight participants in the SW group displayed significant variations in their body times, while the body times of the eight participants in the SWL group were somewhat similar (Figure 6c-e), consistent with the body times of the participants estimated by wristwatch-recorded systolic blood pressures and heart rates (Figure 5 and Supplementary Figure S2).

Discussion
In this study, we integrated multi-day wristwatchrecording, and NanoString-based human body time determination to investigate the effects of the non-24hour military shift on human circadian rhythms and sleep.We found that multi-day wristwatch-recorded physiological parameters such as oxygen saturation levels, systolic blood pressures, diastolic blood pressures, and heart rates display robust rhythmicity (Figures 2 and 3), and the periods, phases, and amplitudes of these physiological parameters can be used to evaluate human circadian functions.Six days of non-24hour military shifts suffice to significantly disrupt the rhythmicity of key physiological parameters of the participants and damaged sleep quality (Figures 5 and 6).In particular, we demonstrated that the administration of 30-minute blue light after waking up in the morning and 30-minute no blue-ray light before sleep in the evening/night improves circadian misalignment and sleep quality, providing an effective and non-invasive approach to improving the circadian functions of non-24-hour shift workers.

Robust rhythmicity of wristwatch-recorded physiological parameters
We recorded six physiological parameters, including oxygen saturation levels, systolic blood pressures, diastolic blood pressures, heart rates, wrist temperature, and steps of the 60 participants via wristwatches for ten consecutive days (Figure 2) and observed that these parameters in the majority of participants all showed robust rhythmicity (Figure 3).The wristwatch method was recently used to monitor the heart rates of 211 volunteers and found that it could reflect the rhythmicity of heart rates and was largely consistent with the rhythmic characteristics of a 24-hour Holter ECG (Zhang et al. 2022).Another study showed the 24-hour amplitude of the wrist temperature of patients with a variety of diseases was significantly lower than that of the control population, indicating that the use of wristwatches can be a good indicator of the amplitude of the circadian clock (Brooks et al. 2023).Our study found that the rhythmicity of the heart rates recorded by the watch is highly correlated with rhythmic characteristics of the systolic blood pressures, diastolic blood pressures, and oxygen saturation, while the correlation with the number of steps is low, indicating that the masking effect of the amount of activity has little effects on the rhythmicity of the heart rates and blood pressures recorded by the wristwatch but affects the number of steps.
Wristwatch-recorded data allows for examining three primary parameters of the circadian clock, i.e. period, phase, and amplitude.The phase reflects the time-of-day-specific peak value of each participant, which is a valuable indicator for the physical health (Dallmann et al. 2014), and the period represents whether people's physiological state is synchronized with the cycle of the external light (Czeisler et al. 1999); while the amplitude can reflect the stability of the circadian clock (Cornelissen and Otsuka 2017).Although the analysis of the period, phase, and amplitude are well used in studying the circadian clock of cell lines and animal models (Maier et al. 2021;Mosser et al. 2019), it has scarcely been used for human studies.The relationships and the differences in the three circadian clock parameters of people of different ages, genders, and latitudes are rarely reported (Zhang et al. 2022).Studies have shown that wristwatch-wearing peoples exercise significantly more than those who do not wear them (Xie et al. 2018).Therefore, these calculated periods, phases, and amplitudes of wristwatchrecorded physiological parameters can remind people to focus on their circadian functions, which should be especially helpful for the shift work population (Phillips et al. 2017).

The non-24-hour military shift work schedule significantly reduces sleep quality and disrupts the circadian rhythms
Salivary melatonin rhythms of 20 crew members were shown to be disrupted during a prolonged six h on duty/ 12 h off duty voyage on a Trident nuclear submarine (Kelly et al. 1999), while all the crew members had adjusted to 18-hour rhythms of body temperature, pulse rates, respiration rates, and systolic and diastolic blood pressures on an 18-h watchkeeping shift submarine (Schaefer et al. 1979), and increased sleepiness and shortened sleep duration for the participants with a simulated 6 h on/6 h off duty watchkeeping system (Eriksen et al. 2006).Our study demonstrated that the non-24-hour military shift work schedule significantly damaged the participants' sleep quality and circadian functions.Six days of the non-24-hour military shifts resulted in reduced sleep duration (Figure 6a,b).Further, analysis of multi-day wristwatch-recorded physiological parameters showed that six days of the non-24-hour military shifts can phase-shift both systolic blood pressures (Figure 5a) and heart rates (Supplementary Figure S2a), reduce the amplitudes of some participants (Figure 5c), and in particular desynchronize these physiological parameters of some participants (Figure 5d,e).Together, these results suggest that six days of the non-24-hour military shifts significantly reduce sleep quality and disrupt the rhythmicity of key physiological parameters of the participants, and provide insights into the extensive damage to human physiology and sleep by the non-24-hour shift work.

Administration of blue light impulse in the morning and no blue-ray light in the evening improves circadian functions and sleep quality
The non-24-hour shifts enforced on military operational units such as submarines have been shown to affect human physiological parameters such as body temperature, blood pressures, heart rates, and melatonin production with reduced sleep time (Kelly et al. 1999;Schaefer et al. 1979;Short et al. 2016).However, none of these previous studies had investigated intervention procedures to improve the impaired circadian functions caused by the non-24-hour military shifts (Guo et al. 2020).We established the optimal lighting apparatus to improve circadian functions and sleep quality (Figure 4).We examined two types of eyeglasses with blue or green wavelengths, and table lamps with two wavelengths (blue light and no blue-ray light), three types of illuminances (strong, medium, and weak).We found that wearing blue eyeglasses for half an hour after waking up in the morning and exposure to a low level (60 lux) of no blue-ray light for half an hour before sleep in the evening can effectively improve the amplitude of the circadian rhythms and sleep quality (Figure 4), and in particular resynchronize the decoupled physiological parameters caused by the non-24-hour shift work (Figure 5f,g).Melatonin is an important sleeppromoting hormone (Cassone et al. 1993).The synthesis of melatonin is not only affected by light but also regulated by the circadian clock (Cassone et al. 1993).Blue light in the daytime can significantly inhibit the melatonin synthesis (Sasseville et al. 2006).Wearing blue light eyeglasses upon waking up inhibits melatonin synthesis, helping get clear-headed as soon as possible (Brown et al. 2022).On the other hand, no blue-ray light at night can promote melatonin synthesis and facilitate sleep (Hester et al. 2021).Future work should investigate if this lighting apparatus can improve patients with circadian rhythm disorders and sleep disorders.Nowadays, various lighting and electronic devices have frequently been used during the nighttime, reducing sleep quality and disrupting circadian rhythms, subsequently leading to numerous sub-healthy problems and even diseases (Esaki et al. 2019;Lai et al. 2020;Lunn et al. 2017).Blue light has long been used to treat patients with mental and sleep disorders (Baglioni et al. 2020;Faulkner et al. 2019) but has rarely been used to alleviate the harmful effects caused by the non-24-hour military shifts.Our findings showed that the administration of blue light in the morning and no blueray light in the evening helps improve the circadian functions and sleep quality of the non-24-hour shift workers.

Estimation of the internal body time by wristwatch-recorded systolic blood pressures and heart rates and the NanoString-based method
We also conducted a correlation analysis of multi-day wristwatch-recorded physiological parameters and found that these parameters were highly correlated (Figure 3).It is reliable to estimate the circadian rhythm of the population by recording and analyzing systolic blood pressures and heart rates.Recording these physiological parameters with wristwatches is non-invasive and not restricted by the location and time of use (Liang et al. 2018;Xie et al. 2018;Zhang et al. 2022).Even though the data recorded with wristwatches are not as accurate as those measured by professional medical devices, the rhythmicity of each physiological parameter from our data and the high consistency between different participants were observed (Figures 2 and 3).While the DLMO method is the gold standard for evaluating the human body time (Klerman et al. 2002), it is timeconsuming, labor-intensive, and expensive, thus not feasible for the large shift work population.The establishment of this method provides an effective means for monitoring the circadian rhythms of regular people.
The NanoString-based method requires only a onetime point at any time point of the day and does not require sampling under dark conditions (Wittenbrink et al. 2018), circumventing some of the shortcomings of the saliva or blood DLMO detection method (Wittenbrink et al. 2018).However, the NanoStringbased method also has certain limitations; for instance, it requires blood samples and a special machine for detection, with a relatively high cost but only for predicting the phase.

Limitations
We selected 60 participants and examined the utilities of wristwatch-recorded oxygen saturation levels, blood pressure, heart rates, wrist temperatures, and steps for computing the period, phase, and amplitude of the human circadian clock in the stage-one experiment.The participants in the stage-two and stage-three experiments were selected from the 60 participants who attended the stage-one experiment.Thus, the participants in the respective groups in the stage-two and stage-three experiments had relatively small sample sizes.All participants are males, as we assume that military operational staff members are mainly malebiased.However, female non-24-hour shift workers are as many as male ones.How females performed in our experimental scheme should be investigated in the future.Moreover, we didn't examine the circadian and sleep parameters of the participants in the recovery time after the stage-three experiment.It will be interesting to investigate whether the SWL participants had a speedy recovery.

Conclusions
Multi-day wristwatch-recorded oxygen saturation levels, blood pressures, heart rates, and other physiological parameters allow for calculating the period, phase, and amplitude of the human circadian clocks, and the phases of systolic blood pressures and heart rates exhibit the highest consistency and can be used as reliable estimators for the human body time.The non-24-hour military shift work schedule significantly disrupts the circadian rhythms and impairs sleep quality.Wearing blue eyeglasses for 30 minutes after awakening, combined with exposure to no blue-ray light for 30 minutes before sleep, improves the stability and synchronization of circadian rhythms.

Figure 1 .
Figure 1.A diagram for the three-stage experiments.(a) Diagrams of stage-one experiment, the stage-two experiments, and the stagethree experiment.(b) The timetables of work, and sleep and leisure of 30 participants in the stage-one + stage-two experiments (top), of the 8 participants in the SW experiment of the stage-one + stage-three experiments (middle), and the other 8 participants in the SWL experiment of the stage-one + stage-three experiments (bottom).Working time in white, and sleep and leisure time in gray.The detail information for stage-one + stage-two experiment (c), stage-one + stage-three in SW experiment (d) and stage-one + stage -three in SWL experiment (e).Pink is working time, purple is sleep and leisure time; centrifuge tube is the number of days for collecting saliva, 8-9 time points are collected every day, with an interval of 2-3 hours; the document icon is the number of days when the questionnaires were collected; glasses indicate the use of blue or green glasses following the wake, and the yellow lamp icon indicates the night light treatment under different wavelengths and intensities.

Figure 2 .
Figure 2. Rhythmicity analysis of multi-day wristwatch-recorded physiological parameters of the 60 participants under the regular work and rest schedule.Phase, period, amplitude, and heatmap analysis of rhythmicity of oxygen saturation levels (a-c), systolic blood pressures (d-f), diastolic blood pressures (g-i), heart rates (j-l), wrist temperature (WT) (m-o), and steps (p-r) of the 60 participants.OS: oxygen saturation; SBP: systolic blood pressures; DBP: diastolic blood pressures; HR: heart rates; WT: wrist temperature.(a, d, g, j, m, p) Each color represents the phase of one representative participant out of the 60 participants.(b, e, h, k, n, q).Each dot represents the period and amplitude for one representative participant out of the 60 participants.(c, f, i, l, o, r) Heatmaps for the six wristwatchrecorded physiological parameters.The horizontal axis is hours, and the vertical axis is participants.For clarity, only 30 participants are shown.P1-P30: participants 1-30.CV: coefficient of variation.

Figure 3 .
Figure 3. Correlation analysis of the participants' six wristwatch-recorded parameters (SBP, DBP, HR, OS, WT, and steps) in the stageone experiment.(a) Rhythmic patterns of six wristwatch-recorded parameters for the P3 and P4 participants in the stage-one experiment.(b) The Pearson correlation panels of SBP, DBP, HR, OS, WT, and steps of P3 and P4 participants in the stage-one experiment.Red is a positive correlation, while blue is a negative correlation.Numbers represent the correlation coefficient r values.*p < 0.05, **p < 0.01, ***p < 0.001.(c) Phase scatter diagram of the six physiological parameters of the 60 participants in the stage-one experiment.The ordinate is the time of the phase (unit is hour).(d) Consistency analysis of the period, phase, and amplitude of the five wristwatch-recorded parameters of the 60 participants in the stage-one experiment.SBP, systolic blood pressure; DBP, diastolic blood pressure; HR, heart rate; OS, oxygen saturation; WT, wrist temperature.P3: participant three; P4: participant four.

Figure 4 .
Figure 4.The spectrums of the lighting apparatuses used in the stage-two experiment and their effect on the circadian rhythms and sleep characteristics of the six groups.(a) The spectrums and photos of the blue light glasses (top), green glasses (middle), and sleeppromoting lamps (bottom).(b) Comparison of the period and phase of the systolic blood pressures of the representative participants of the six groups in the stage-two experiment with those of the same participants in the stage-one experiment.(c) The effects of the lighting apparatus on the amplitudes of systolic blood pressures of the six groups.(d,e) Comparison of total sleep time (d) and sleep latency (e) of the representative participants of the six groups between the stage-one experiment and the stage-two experiment."Pre" represents the stage-one experiment, and "after" represents the stage-two experiment.CV: coefficient of variation.

Figure 5 .
Figure 5.The impact of the optimal lighting apparatus on the systolic blood pressures of the SW and SWL groups in the stage-three experiment.(a,b) Comparison of heatmaps, periods, and phases of the systolic blood pressure rhythms of the eight participants of SW (a) and SWL (b) in the stage-three and stage-one experiments.(c) Statistical analysis of the amplitude, period, and phase of the systolic blood pressures in the stage-three and stage-one experiments.(d-g) Correlation analysis of phases of systolic blood pressures, diastolic blood pressures, heart rates, and oxygen saturation levels of the 8 participants of SW (f) and SWL (g) groups between the stage-one experiment (d) and the stage-three experiment (e).SW: non-24-hour shift work group; SWL: non-24-hour shift work group treated with the optimal lighting apparatus.SBP, systolic blood pressure; HR: heart rate; pre or −1, stage-one experiment; after or −3, stage-three experiment.Red is a positive correlation, while blue is a negative correlation.Numbers represent the correlation coefficient r values.*p < 0.05, **p < 0.01, ***p < 0.001.CV: coefficient of variation.

Figure 6 .
Figure 6.The effect of the optimal lighting apparatus on sleep characteristics and NanoString analysis of peripheral blood lymphocytes of SW and SWL groups in the stage-three experiment.(a) Comparison of the total sleep time of the participants between the SW and SWL groups in the stage-three experiment.(b) Comparison of the difference in total sleep time of the participants of the SW and SWL groups between the stage-one experiment and the stage-three experiment.(c) The average chronotype comparison between the SW and SWL groups before and after the non-24-hour shift work schedule.Each dot represents the body time of one participant.The paired analysis for SW (d) and SWL (e) before and after non-24-hour shift work schedule.SW: non-24-hour shift work group; SWL: non-24-hour shift work group treated with the optimal lighting apparatus.Pre: stage-one experiment; after: stage-three experiment; after-pre: the difference between the stage-three experiment and the stage-one experiment; DPS: day post-shift work schedule.