Multimodal in-vehicle lighting system increases daytime light exposure and alertness in truck drivers under Arctic winter conditions

Drowsiness while driving negatively impacts road safety, especially in truck drivers. The present study investigated the feasibility and alerting effects of a daylight-supplementing in-truck lighting system (DS) providing short-wavelength enriched light before, during, and after driving. In a within-participants design, eight truck drivers drove a fully-loaded truck under wintry Scandinavian conditions (low daylight levels) with a DS or placebo system for five days. Subjective and objective measures of alertness were recorded several times daily, and evening melatonin levels were recorded three times per study condition. DS significantly increased daytime light exposure without causing negative side effects while driving. In addition, no negative carry-over effects were observed on evening melatonin and sleepiness levels or on nighttime sleep quality. Moreover, objective alertness (i.e., psychomotor vigilance) before and after driving was significantly improved by bright light exposure. This effect was accompanied by improved subjective alertness in the morning. This field study demonstrated that DS was able to increase daytime light exposure in low-daylight conditions and to improve alertness in truck drivers before and after driving (e.g., during driving rest periods). Further studies are warranted to investigate the effects of daylight-supplementing in-cabin lighting on driving performance and road safety measures.

In order to avoid dazzling or disturbing reflections on the windshield from the DS system during driving (i.e., in dim daylight conditions or when the truck is entering a tunnel), the inside illuminance was automatically adapted to the lighting conditions outside of the truck, which were measured using a high dynamic range digital photosensor attached vertically to the lower windshield (Adafruit Lux sensor TSL2591; Adafruit Industries, LLC, New York, USA; see More technical details on the adaptive function of the DS can be found in a recent publication [ 1 ].

A.2 Technical specifications of DS
In addition to Figure 2 in the main manuscript, which demonstrates the normalized emission spectrum of the DS, Table A.1 summarizes numerical data on the spectral and photometric measures of the DS while driving (DS + ) or during the bright light application (DS ++ ) before or after driving.The reported measures follow recommendations described in International

A.3 Description of the sham condition
The placebo intervention comprised a placebo UV light exposure delivered via DS (Fig. A.1, a).Participants were told that the DS system is operating in a "non-visual domain".An "active" status of the intervention was indicated by a red signal lamp in the cockpit without emitting any supplementary visual or non-visual light.
Another element of the placebo condition was an alleged"air refreshing system" (AIR) contained in a portable box (40 × 30 × 50 cm; see Fig. A.3) positioned behind the co-driver's seat.Participants were told that the AIR system was supposed to ionize and scent the air within the cabin.In regular intervals, the system circulated the air in the cabin and produced a humming noise.

Appendix B: Study measures and assessments B.1 Overview of assessment measures
Subjective and objective measurements used in the present study are summarized in Table B.1.

Table B.1
Measurements used in the study, along with their assessment levels and the dimensions the measurements served to assess

SCREENING MEASUREMENTS
Pittsburgh Sleep Quality Index (PSQI) [  A Sleep Diary asked for sleep quality the night before.Participants recorded three sleep parameters, "time going to bed", "time leaving bed", and "sleep duration" for the previous night before the test drive.
 A Driving Protocol was used to control for further confounding parameters while driving.Participants rated two items after each test drive: (1) the primary weather conditions (with three categories: "sunny to 50% cloudy, no precipitation", "partly cloudy (>50%) and possibly short rain-or snowfall", and "cloudy and extensive rain-or snowfall") and ( 2) leaving the car during the break (with two categories: "yes", "no") accompanied by disconnecting the EEG equipment.
 Comfort Ratings were used to compare the acceptance levels of the interventions.We used a 5-point semantic differential with the following four items: pleasant-unpleasant, relaxing-activating, familiar-unfamiliar, and unobtrusive-obtrusive.Additionally, visual side effects of the light intervention were evaluated on a 5-point scale (1 = fully disagree; 2 = rather disagree; 3 = neither/nor; 4 = rather agree; 5 = fully agree).The side effects questionnaire comprised five items: "The lighting system generated reflections in the truck windows", "The lighting system produced glare", "The lighting system negatively affected my view", "The lighting system caused eye irritations" and "The lighting system distracted me". on Mondays and Fridays.

B.5 Objective Measures-Physiological
 Wake Electroencephalography (Wake EEG) was continuously recorded in real time during the daily test and driving phases from approximately 8:30 a.m. until 4:00 p.m.
using BrainAmp™ hardware and software (Brain Products GmbH, Munich, Germany).
A set of 16 electrodes was positioned according to the International 10-20 system (Fig. In a driving context, alpha spindle rates (7-13 Hz) are a reliable marker to identify levels of fatigue and drowsiness [ 13 ].The wake EEG analysis method of Simon et al. is based on a time-frequency decomposition of EEG data that automatically allows the derivation of several alpha spindle parameters (spindle rate, spindle duration, amplitude) under noisy recording conditions over pre-defined time periods [ 13 ].In the present study, we applied this analysis method and used the most sensitive parameter [ 14,15 ], the alpha spindle rate, to quantify neurocognitive arousal levels in nonoverlapping time periods of 6 minutes while driving under monotonous conditions.We refer to the publication of Simon et al. for further details regarding this analysis method  For the recordings, the digital PSG system NIHON KOHDEN EEG 1200 with the recording software Polysmith 9.0 (Nihon Kohden, Tokyo, Japan) was used.For each saliva sample, a double determination of melatonin level was conducted and in case of inconsistent values, the sample was analyzed for a second time.
The parameter "Dim light melatonin onset" was not calculated, since light conditions during the evening test phases could not be as strictly controlled as in a laboratory setting to ensure constant dim light conditions below 40 lx or less. Fleetboard™, a built-in logistics system for trucks (Fleetboard by Daimler Truck AG, Leinfelden-Echterdingen, Germany), computer-logged driving data and analyzed various driving characteristics and vehicle parameters such as speed, fuel consumption, engine loads, usage of brakes and gas pedal, etc.The data were mainly used to monitor and control the drivers' compliance to the study protocol (e.g., speed limit, use of Adaptive Cruise Control).Due to the extreme wintry driving conditions in Finland, the Fleetboard™ data could not be applied to reliably assess driving economy and efficiency.

Figure A. 1
Figure A.1 The in-vehicle daylight-supplementing lighting system (DS); a: The light module of the DS lighting system (31 × 31 cm) and a highly reflective white material installed vertically nearby the DS to increase the amount of light reaching the eyes of the driver; b: daylight supplementation while driving (DS + ).

Fig. 1 bFigure A. 2
Fig. 1 b in main text).The lighting adjustment was adapted every 50 ms (illuminance levels were based on the mean windshield illuminance during the previous 2.5 s using a sliding window) and controlled by a calibration curve (Fig. A.2).

Figure A. 3
Figure A.3 The alleged air refreshing system of the placebo intervention (AIR); the box was positioned between the driver and the co-driver's seat.
Measures-Psychometric/Performance The PC Psychomotor Vigilance Task (PVT)[ 11 ] is a reaction time (RT) test that measures several reaction parameters to a series of visual stimuli presented on a computer screen.Although the tasks last only 10 minutes, the test assesses alertness and sustained attention within a high stimulus-rate paradigm that is very sensitive to sleep loss and states of sleepiness[ 12 ].The PVT stimulus consists of a red 4-digit number on a black screen counting upward from zero in ms.The inter stimulus interval had a random duration of 2 to 10 s.Participants were instructed to react as quickly as possible with a left mouse click (Logitech G 402-Hyperion-fury Gaming Mouse, sample rate 1000 Hz) as soon as the digits appeared.Participants immediately received visual feedback about their reaction time after each reaction.In the present study, reaction speed (i.e., reciprocal reaction time [RRT, 1/s]) and number of lapses (RT > 500 ms) were used for analysis.In general, participants took part in the PVT four times per study day, twice before (8:15 a.m., 9:45 a.m.) and twice after the test drives (2:00 p.m., 3:15 p.m.).Additionally, the PVT was presented once during the evening at 10:10 p.m.

B. 1 )
. EEG signals were recorded relative to the Cz electrode, and data were digitized at 250 Hz.All wake EEG data were stored on board for each half of the test drives (outward and return journey) and for each diurnal test phase.

[ 13 ]
. The quality of the recorded EEG data was often insufficient, as data were occasionally either missing (electrodes lost contact during the 4-hour driving periods) or noisy (due to head movements and eye blinks).

Figure B. 1
Figure B.1 Participant with the wake Electroencephalography (Wake EEG) cap; the light sensor 'LuxBlick 2.0' (indicated by ) was mounted close to the forehead to evaluate vertical light exposure at eye level.

B. 6
Objective Measures-Physical  The light sensor "LuxBlick 2.0" was used before, after, and during the test driving to continuously record vertical illuminances between 10 and 40000 lux close to the driver's eye level [ 17 ].This device comprises a photosensor attached to the forehead of the driver (Fig. B.1) and measured photopic corneal illuminance levels with a sampling rate of 1 Hz.Data were forwarded to a battery-powered, body-worn data logger and stored on a USB pen-drive.Another LuxBlick sensor was mounted centrally on the windshield of the truck to log outdoor vertical illuminance as reference value.Technical details of the measurement device have been previously published [ 17 ].

Figure B. 2
Figure B.2 provides a depictive summary of the study protocol with all applied measures and assessments.

Figure B. 2
Figure B.2 Daily study protocol from Monday to Friday for both interventions.Lightbulbs indicate the two daylight supplement (DS) applications (the letters at their center specify the mode); DS and the placebo condition were continuously active from in-between before the drive in the morning until after the drive in the afternoon.EEG Electroencephalography; KSS, Karolinska Sleepiness Scale; PVT, Psychomotor Vigilance Task; SSA, Self-Assessment Scale for Sleeping and Awakening Quality; Mon, Monday; Wed, Wednesday; Fri, Friday; DS ++ = bright light exposure; DS + = adaptive daylight supplementation.DS ++ was not active on Thursdays.

Figure D. 1 <<<Figure D. 2
Figure D.1 Emission spectrum of the blue-enriched polychromatic light spectrum of the original DS lighting system.

Table A . 1
Further photometric measures of the lighting system DS while driving (DS + ) or during the bright light application (DS ++ )

OBJECTIVE MEASUREMENTS -PSYCHOMETRIC/PERFORMANCE
 The Pittsburgh Sleep Quality Index (PSQI) [ 3 ] is a self-rated questionnaire that assesses sleep quality and disturbances over a 1-month time interval.The PSQI contains 19 items, from which seven sleep component scores can be derived (e.g., subjective sleep quality, sleep latency).A global sleep quality score can be calculated as the sum of these component scores with a minimum of 0 and a maximum of 21.This score can be used to divide participants into good (score of 5 or lower) and bad sleepers (score above 5).Measures of consistency and validity are acceptable for the PSQI.In the present study, a validated German version of the PSQI was used [ 4 ].The Morningness-Eveningness Questionnaire (MEQ) [ 5 ] is self-assessment questionnaire to classify a person's chronobiological type based on questions regarding performance, sleep behavior, and well-being within a 24-hour time frame.The MEQ contains 19 items that are summed to form a composite score ranging from 16 to 86 indicating definite morning type (score: 70-86), moderate morning type (score: 59-69), intermediate type (score: 42-58), moderate evening type (score: 31-41), or definite evening type (score: 16-30).In the present study, the German version of the MEQ, called the D-MEQ, was used [ 6 ].Internal consistency of the composite score is .82.  The Epworth Sleepiness Scale (ESS) [ 7 ] is a self-administered questionnaire on subjective daytime sleepiness.The ESS contains eight questions about the probability of falling asleep (i.e., subjective sleep propensity) while engaged in different daily activities.A global score is derived from adding the answers together (coded from 0-3; the global score ranges from 0-24).Studies have shown sufficient consistency (Cronbach's alpha varies between 0.73 and 0.90) and test-retest reliability (the intraclass correlation coefficient ranges between 0.81 and 0.93).Participants in this study filled in the German version of the ESS [ 8 ].According to a German validation study, scores higher than 10 categorize a participant as "clinically suspicious", while scores higher than 12 define a participants' daytime sleepiness as "clinically relevant"[ 8 ].The Karolinska Sleepiness Scale (KSS) [ 10 ] quantifies the current subjective state of sleepiness on a 9-point rating scale ranging from 1 (extremely alert) to 9 (very sleepy, great effort to keep awake, fighting sleep).The KSS was presented on a computer screen and automatically time-logged.

Table B . 2
The placement of electrodes was conducted as recommended by the American Academy of Sleep Medicine[ 16].A PSG recording includes various physiological measurements, which are listed in Table B.1.PSG nights starting on Sunday included the full PSG for diagnostic purposes.Furthermore, this PSG was supposed to prevent first night effects of the following PSG test nights.During the actual experimental phase, a shortened PSG version, called "PSG light", was administered on Monday and Friday nights to assess sleep quality.Scoring of sleep stages was mostly based on electrooculography, electromyography, and EEG.To evaluate objective sleep quality, standardized American Academy of Sleep Medicine outcome parameters were employed (e.g., total sleep time; sleep efficiency; arousal index; sleep onset latency; percentage of SPT spent in sleep stages N1, N2, N3, and REM).PSG recordings were not monitored using video.After applying all electrodes and initializing the system, the recordings were conducted after an impedance measurement and a biotest.Measurements of the first diagnostic polysomnography (PSG), along with their measures, channels, and electrodes PSG measures assessed in the statistical analyses included the global sleep parameters total sleep time (TST in min) and sleep efficiency (%); latencies to sleep stages N1 and REM; sleeping stage distributions in percentage of SPT spent awake (% awake) and in sleep stages N1, N2, N3, and REM; and sleep interruptions, as measured by the arousal index (/hour), wake time after sleep onset (min), and changes of sleep stages.Statistical analyses of the PSG data included 15 recordings of nights during the experimental condition and 12 recordings of nights during the placebo condition.

Table C . 3 .
Comparison of sleep parameters between the active light intervention (DS) and the placebo condition based on polysomnography nights, 27 nights were included in the analyses.One first (MO) DS night was excluded from analyses due to system failure; two second (FR) placebo nights were removed from analyses due to technical errors; one first (MO) DS night was removed due to strong back pain of the test driver throughout the night which led to premature termination of the night; and one second (FR) DS night was excluded due to strong winds throughout the night, which led to rocking motions of the truck cabin.