Exploring the effects of spectral light exposure on University students' sleep quality: a cross-sectional study

. In developed nations, 2/3 of the population fail to obtain the recommended 7-9 hours of sleep, with large consequences for health and productivity. A potent stimulus in regulating sleep and wake balance is light received at a light-sensitive retinal protein called melanopsin, found in our retinal cells, otherwise known as intrinsically retinal ganglion cells (ipRGCs). Hence, to better understand the effects of bedroom indoor environmental quality IEQ on sleep, we set up a study to objectively explore the impact of spectral light exposure through non-visual pathways on the participants' sleep quality. For one week, University students measured IEQ in their bedrooms while monitoring their spectral light exposure during the day and sleep quality at night. The students were asked to keep the bedroom environment unchanged for the measurement period. Our preliminary results show trends between 7-day accumulative light exposure and length of deep sleep. This trend is not significant, and further in-depth analysis is needed. Among other limitations in the processed data, the diverse and mixed demographic of the sample was not considered. The students' living conditions (single vs. family, accommodation, etc.) varied in this initial analysis.


Introduction
Exposure to the right amount of light has an undeniable effect on occupants' well-being, with impacts on health [1], performance [2], and sleep [3]. Regular exposure to sunlight has long-term beneficial physical and mental health effects. The reason lies in the dual function of the human visual system. The visual system enables us to see and resets the internal body clock. Fig.1 depicts the two distinct pathways that light and daylight affect our visual environment, mood, and sleep. Visual responses in humans are mediated mainly by the rods and cones. In contrast, non-visual responses can primarily be driven by the intrinsically photosensitive retinal ganglion cells (ipRGCs) [4], which serve as an essential input in synchronizing the human circadian rhythms to the daily 24-hour cycle of light and dark. The ipRGCs can capture light separately from the circuits of cone and rod photoreceptors used for vision. However, all photoreceptors play a role in visual and non-visual pathways [4]. The non-visual responses include physiological and neurobehavioral responses [5], such as melatonin suppression, sleep quality, temperature fluctuations of the core body, and alertness, which are directly or indirectly essential to human health, cognition, and performance [6]. The advances and delays of the circadian rhythm can disrupt the aforementioned biological functions.
Daylight naturally creates the cycle and the amplitude of the light and dark ratio, which is essential for circadian entrainment. Humans are diurnal and associate their wakefulness and activity with lit hours, while sleep and rest cycles are commonly tied to darkness [7]. Therefore, a high light contrast between day and night is required to maintain optimal circadian rhythms [3]. Moreover, circadian triggers and stimulation occur in the presence of five key indicators of spectral lighting: timing, intensity, duration, spectrum, and previous exposure to light [8]. In other words, depending on the aforementioned key indicators, circadian stimulation could differ with consequences on our natural biological functions.
One biological function highly dependent on our circadian rhythm is sleep and sleep quality. We spend 90% of our time indoors with excessive exposure to lower intensities, a limited spectral range of artificial light, and lack of daylight, which is the recipe for circadian disruption. The disruption to the sleep-wake pattern has a potential role in brain defects, sleep reduction, and mental health, with consequences on productivity and health issues such as the increased risk for cancer [9]. Moreover, securing the indoor environment quality (IEQ) and occupants' well-being while indoors requires better of their daily light exposure. To objectively explore the impact of light exposure on the participants' sleep quality in relation to their bedroom IEQ [10], we set up an explorative crosssectional study involving University students. The study was extensive regarding data acquisition and physical measurements with acute and accumulated light exposure. Here we present some preliminary results. More in-depth analysis and data acquisition is ongoing.

Circadian stimulation key indicators
Human daily cycles are induced by key indicators such as timing, spectrum, light intensity, and duration of light exposure [11]. Understanding the key indicators of circadian stimulation is necessary to identify circadian triggers that could affect sleep patterns.
The non-visual responses adapt to light intensity and spectral composition variations over a more extended period than the visual system. Relatively low light levels can be enough to operate specific visual tasks but are not necessarily enough to sustain the operation of circadian systems or induce other physiological and behavioral non-visual responses [12]. Past light exposure can stimulate circadian responses, extending over several hours or even days [13]. Light exposure in the early morning can shift circadian rhythms forward, whereas insufficient illumination, especially if paired with inappropriate light exposure in the late afternoon, will drift the circadian phase. Moreover, regarding the spectral light intensity within a wavelength region from about 380 to 800 nm, brighter bluish light at a shorter wavelength significantly influences human biological response, enhancing alerting effects and cognitive function than dimmer yellowish light at a longer wavelength [14].
Timing. The internal clock has a period-limited sensitivity to light. It is known as phase delay and phase advance. Light exposure in the morning is more effective at resetting the circadian system than in the evening.
Light Intensity and Duration. The intensity also affects the circadian system. The circadian rhythm is better regulated by bright light than by dim light. In addition, the duration of light exposure also influences the circadian system. Extended periods of light exposure are more effective at resetting the circadian rhythm than shorter periods. However, there is a non-linear relationship between intensity, duration, and the magnitude of circadian rhythm resetting.
Spectrum. It is also essential to consider wavelength, as ipRGCs signal through melanopsin, whose spectral sensitivity is primarily bluish. It is why shortwavelength light is most effective in stimulating the circadian system. Long-wavelength light gives, on the other hand, minimal circadian stimulus and therefore suppresses only some melatonin. Long-wavelength light is, therefore, helpful in promoting alertness in the late afternoon and evening without disrupting the circadian system.
History. The circadian system shows adaptation to previous light exposure in determining response to a current stimulus.

Circadian metrics
Developed to meet our visual needs, standards, and practices within daylight design are ahead of our scientific understanding of how light regulates various bodily functions. ipRGCs are unique because they can send signals through a photopigment called melanopsin. Melanopsin is light-sensitive to short wavelengths peaking in the blue region of the visible spectrum (490 nm). In contrast, the visual system, mediated by photoreceptors rod and cone cells, is most sensitive to wavelengths peaking in the green region of the visible spectrum (555 nm). Thus, the description of circadianeffective daylight needs to be more adequate in terms of quantities and units dependent on the visual system [12]. Recent research shows that the visual quantity' photopic illuminance' is inappropriate for evaluating non-visual effects. Several researchers have developed various metrics to address this issue. Table 1 lists the most prevalent spectral circadian metrics. The spectral distribution of light in circadian metrics is generally weighted by the spectral circadian efficient function of curves. Among the most well-known and cited are namely Circadian Stimulus (CS) Rea et al. [15], Equivalent Melanopic Lux (EML or mlux used in the paper interchangeably) Lucas et al. [16], and Melanopic Equivalent Daylight D65 Illuminance (mel-EDI) [17]. The two latter prevalent methods to quantify the light stimulus are based on the spectral response of photopigment in the rods, cones, and ipRGCs and nocturnal suppression of the melatonin hormone. Lucas et al. introduced EML, which is calculated as the ratio of the melanopsin-activating radiation of a light source to its photopic-activating radiation multiplied by 1.218. Since EML is not recognized by the International System of Units (SI), CIE has proposed Melanopic Equivalent Daylight Illuminance [18]. Melanopic EDI is the Illuminance of standard daylight at 6500K at a point that provides the same melanopic irradiance as the test source. Both methods require two inputs, the spectral power distribution of a light source (SPD) and the photopic Illuminance at the plane of the eye. These methods of Lucas et al. and CIE are based solely on photopigment signals, while CS is expressed as a decimal percentage of the nocturnal melatonin suppression. Quantitative design thresholds for circadian lighting design are recommended by the WELL Building Standard (WELL) [19]. The WELL thresholds mainly rely on EML while allowing compliance with mel-EDI. Furthermore, Brown et al. defined a list of recommendations for different exposure and threshold during morning and evening for enhanced circadian stimulation indoors [17].

Methodology
A cross-sectional study was performed from February to May 2022 in the Greater Copenhagen areas in Denmark. 96 college-age students attending the "Indoor Climate" course" were invited to a one-week-long measurements and data acquisition as part of their course assignment.

Experimental Procedure
For one week, the University student participants monitored their exposure to light in the visible spectrum. Meanwhile, their bedrooms' IEQ and sleep quality were also measured. The students were asked to keep the bedroom environment unchanged for the measurement period. Students filled sleep diaries and conducted a 3minute Baddeley performance test integrated into each sleep diary's end on selected evenings and mornings. The daily light exposure was measured during the day using a wearable RGB-sensor Fig3. The sleep quality of each person was tracked using Fitbit-actiwatch, which they wore at all times. Each student was asked to attend an instruction session prior to doing the measurements and received all the necessary instruments. The students were instructed to do the measurements and receive a checklist to ensure maximum coherence in the measurement procedures. All measurement instruments were gathered and organized in blue boxes Fig 2. The students were organized in groups of four as part of their course curriculum. Each group received two boxes.
The instrument "BOX, was used in the first week (from Wednesday to Wednesday) by two students and then passed on to the other two students in each group. The switch occurred at the University once all measuring data was was collected.

Physical Measurement
The measurement instruments were placed in the blue instrument, the "BOX, " Fig 2. The following instruments were used in the measurements. The specification of the sensors are listed in Table 2. HOBO loggers: temperature, humidity CO2 monitor connected to the logger, Flow sensor: PM2.5, PM10, TVOC, and NO2, LYS sensor: RGB light exposure, Ibutton: skin temperature, Fitbit actiwatch for sleep quality, Extension wires and all necessary chargers,

Measurement and Calculation of EML
The spectral light was measured using an RGB color sensor called LYS. This device allows for the registration of light's spectral composition, mimicking the human eye's trichromatic vision capability. The sensors are designed to be installed on the clothing to measure the exposure to light as close as the reception of light at the eye. The logged data is stored on the Cloud and is accessible through an online dashboard.  process in the LYS technology protocol calculates and reports the RGB values, the Illuminance at the eye level, Correlated Colour Temperature (CCT), and movement.
In this study, we only used melanopic lux calculated by LYS and weighted by Lucas spectral circadian efficiency function.

.Measurements of sleep quality
Fitbit sleep trackers were used to track sleep quality. The trackers have been reported to be in good agreement with polysomnography in measuring time asleep (total sleep time) and sleep efficiency. They provide gross estimates of sleep stages, although they have been reported to underestimate sleep latency [20]. The actiwatches were worn at all times and during sleep. At the start of each measurement session, the watches had to be synchronized using an App on the tablet to ensure that the recorded data was uploaded to the Cloud. The trackers log the duration of sleep (D_sleep), the number of awakenings (N_awake), minutes awake (M_awake), minutes of deep sleep (Deep), and minutes of rapid eye movement (REM) sleep, and time in bed (TIB) among others. The sleep quality of the participants was determined by defining sleep efficiency and percentages of light, deep, and REM sleep. We calculated the sleep efficiency was calculated by "(minute's asleep ÷ TIB) × 100". The other sleep quality indicators were calculated by "(minute's light/deep/REM sleep ÷ minutes asleep) × 100" [10]. The higher percentages of Deep and REM sleep, alongside a lower percentage of light sleep and higher sleep efficiency, indicate better sleep quality.

Results
Eighty-one data points were retrieved with no faults in data acquisition, logging, data processing, or other external issues. As a first step, we explored the melanopic lux values concerning sleep duration and quality. The key indicators for circadian stimulation, as mentioned in section 1.1, are not only the exposure to a specific spectrum but also duration, timing, history, and intensity. In order to have a better understanding of the effects of light exposure on sleep and to include the key indicators of circadian stimulation, we extracted the EML levels above 200 lux and in durations larger than 10 minutes of exposure for each day [19]. We calculated the average EML levels for each >10 min duration and the rate of exposure to such stimulating intensities.

Accumulative EML exposure
The exposure behavior among the participants varied extensively. Figures 4a and 4b compare the average daily EML exposure of two participants during their data acquisition. The size and color of each data point show the duration of exposure to the average mlux on each day. Participant 4a is exposed each day to mlux between 300 to 1500. It can be noticed that the participants had been exposed to higher intensity levels and has been exposed to daylight. Participant 4b, on the other hand, is exposed to lower ranges of intensity during the week, indicating less exposure to sunlight. This wide range of variation is also visible in Figure 5, where the daily exposure of all participants over time is shown. Fig. 5 shows the accumulative light exposure of all participants throughout the study. There is a general trend showing the participants were exposed to higher light levels as the days progressed from winter to springtime. The median EML is 364 EML in durations lower than 30 minutes.

Sleep data
The participants' sleep quality was calculated based on sleep efficiency and light, deep, and REM sleep percentages. Fig. 7 shows the sleep quality indicators for all participants. The first quality indicator, sleep efficiency, across all subjects shows a median of 88%. However, the higher percentage of light sleep and a lower percentage of deep and REM could indicate that the sleep quality of the participants was not optimal.

A Correlation study
In a correlation study, we explored the relation between exposures to EML above 200 and sleep. In this first analysis, we did not check the time of the day for each exposure, and we did not distinguish between exposure to daylight or artificial light. The accumulative EML intensity, weighted EML by the exposure length, and rate of exposure during the 7-day data acquisition was correlated with the Fitbit sleep data. Table 3 shows the R squared for Pearson's correlation by transforming the correlation to create a t-statistic with the number of Observations minus 2 degrees of freedom.  Concretely, the correlation study is between EML intensity, weighted EML by the duration of exposure, the length of exposure to EML above the 200-lux threshold, and the rate of exposure to EML above the 200-lux threshold. The aforementioned are correlated with duration of sleep, minutes of awake, number of awake, REM, and deep sleep. We do not see any correlation between the mentioned parameters. There is a slight tendency between the intensity of exposure and deep sleep but with low significance. The P values were all above 0.1.

Conclusion
In a cross-sectional study on bedroom IEQ and sleep quality, we invited University students to perform comprehensive measurements of their bedroom quality, sleep, and performance as part of a study course assignment. The students performed physical measurements, subjective assessments, and performance tests throughout the week. The data acquisition occurred over three months. As a first step, we performed a preliminary explorative analysis focusing on light exposure and sleep quality. In this step, we only used the data obtained using a wearable sleep tracker, Fitbit, and a wearable RGB sensor, LYS. The measured IEQ parameters were not processed in this step. In the analysis, we did not take into account the type of housing, i.e., apartments, villa, student dorm, nor the living conditions, i.e., single vs. family accommodation, where the bedroom measurements occur. The diverse and mixed demographic of the sample was not considered in this initial analysis. The preliminary results show that exposure to higher intensity of weighted light, i.e., EML, results in higher minutes of deep sleep. These effects are insignificant, and further in-depth analysis is needed to explore this finding. The students were generally exposed to only shorter periods of higher light intensities. The overall sleep quality also shows that this sample of students were having a lower percentage of deep sleep and REM. In future work, we will overcome the mentioned limitations in our data processing. Exposure to light gradually increased from wintertime to late spring; hence seasonal changes could create a nuisance in the data. We will check the seasonal effects and time of the day for each exposure. Finally, we will try to distinguish between exposure to daylight and artificial light, and we will look into exposure to different light channels of RGB.