Contribution of Electrophysiological Signals in the Study of Thermal Comfort: a Case Study

Thermal comfort is defined as a mental condition that expresses satisfaction with thermal environments, therefore being subjectively assessed by the individual. Nonetheless, thermal comfort (or discomfort) causes a physiological response that may be translated objectively by electrophysiological signals. Additionally, the sensation of thermal comfort also depends on ambient conditions, such as air temperature and relative humidity, which can be easily quantified. In this way, the present study aimed to assess the physiological response in different thermal comfort environments by collecting and analyzing objective parameters of electrical brain and cardiac activities, as well as thermal comfort parameters. Herein, we enrolled five healthy participants, 2 females and 3 males, aged between 22 and 27 years, and with clothing insulation levels between 0.26 clo and 0.52 clo. These participants underwent a mental task whilst exposed to two different thermal environments - one considered to be warm and the other comfortable. Electroencephalography and photoplethysmography techniques were used to collect physiological data from which brainwave activity and heart rate parameters, were respectively derived, and correlated with thermal comfort indexes as computed from the measured air temperature, relative humidity and other thermal quantities. Our results seem to indicate the existence of a relationship between electrophysiological response and thermal comfort indexes, in accordance with the literature. In thermally uncomfortable environments, brain Delta (0.5-4 Hz) activity was observed to be higher, whilst the Beta (13-30 Hz) activity was lower, as well as heart rate values were in general observed to be higher and heart rate variability values lower in comparison with thermally comfortable environments. Whilst these results were promising, further research should be performed to validate our findings, including a larger sample size and more variate thermal environments.

(0.5-4 Hz) activity was observed to be higher, whilst the Beta (13-30 Hz) activity was lower, as well as heart rate values were in general observed to be higher and heart rate variability values lower in comparison with thermally comfortable environments.Whilst these results were promising, further research should be performed to validate our findings, including a larger sample size and more variate thermal environments.

INTRODUCTION
Every day each person may be exposed to different thermal environments, which may lead to changes in thermal comfort.Several authors define thermal comfort as the "mental condition in which the individual expresses satisfaction with the thermal environment that surrounds him/her" and which is subjectively assessed by the individual [1]- [3] .This perception of thermal comfort is determined by heat transfer processes that occur between the person and the surrounding environment, dependent on environmental factors, such as temperature, air speed or relative humidity, among others, as well as personal factors, such as the level of thermal insulation provided by clothing and the person's metabolic rate.
The evaluation of thermal comfort is made through thermal comfort indexes, namely, the Predicted Mean Vote (PMV) index, which allows estimating the average value of the votes of a group of people about the thermal sensation felt, and the Predicted Percentage Dissatisfied (PPD) index, which predicts the percentage of people dissatisfied with the conditions of a given environment.The environment can be considered comfortable when the PMV index is between -0.5 and +0.5 and when the PPD index is less than 10% [ASHRAE 55 or ISO 7730] [4].These indexes were proposed by Fanger in 1970 [5].Another index that is also used as an indicator of thermal comfort is the Wet Bulb Globe Temperature (WBGT).This index measures the thermal stress of a person inserted in a hot environment, which when high, leads the individual to a stress situation, whereas in thermal comfort conditions lies between 20ºC and 27º [2].When environmental conditions are not favorable, the body triggers thermoregulation responses which, in extreme situations, may result in serious physiological and behavioral problems, namely neurological, cardiovascular, and respiratory problems, dehydration, fatigue, stress and anxiety, among others.
As the human being is homeothermic, its internal temperature is kept constant through the thermoregulatory system which, in turn, is coordinated by the central nervous system.In this way it is possible to study the response of the thermoregulatory system to a situation of thermal discomfort through the electrical brain activity, evaluated by the electroencephalographic signal.It is also of interest to evaluate the electrical cardiac activity by means of the electrocardiographic signal, obtained directly or indirectly by photoplethysmography (PPG), studying quantities such as heart rate (HR) or heart rate variability (HRV).
Electroencephalography (EEG) is a technique of electrophysiological monitoring which registers the electrical activity of the brain, and which signal is decomposed in different frequency bands: Delta waves from 0.5 Hz to 4 Hz; Theta waves from 4 Hz to 8 Hz; Alpha waves from 8 Hz to 13 Hz; Beta waves from 13 Hz to 30 Hz; Gamma waves above 30 Hz [6], [7].According to several authors, the Delta activity tends to be higher with increasing levels of thermal discomfort [6], [8], [9].As for the Theta activity, the literature states that it is higher in thermally comfortable environments [8].Regarding Alpha, Beta and Gamma waves, some studies have concluded that their activities are higher in environments considered as thermally comfortable [6], [10][11][12][13] PPG is an optical technique that detects changes in blood volume in the microvascular tissue under the skin and is commonly used for HR and peripheral oxygen saturation (SpO2) [14].Thus, despite being an indirect measurement technique, it is possible to estimate HR and HRV from it.The normal HR of an adult at rest varies between 60 to 100 bpm.In addition to each person's physical activity, HR can be affected by other factors such as temperature and relative humidity.When environmental conditions are unfavorable, the heart pumps more blood leading to an increase in HR [15], [16].On the other hand, HRV corresponds to the oscillations between heartbeat intervals.Like HR, HRV also responds to stimuli, both physiological and environmental [17].Therefore, lower values of HRV are associated to stressful situations and higher values are generally associated to moments of deep breathing and/or meditation [18].
Although this topic can be found in the literature as is summarized in Table 1, to the best of our knowledge there is no prior research that studied this set of physiological variables and simultaneously evaluated objectively the thermal comfort of the environment by the computation of thermal comfort indexes.Therefore, this work aims to study the physiological response in different thermal comfort environments by collecting and analyzing objective parameters: activity of brainwaves, HR and HRV as electrophysiological parameters, as well as PMV, PPD and WBGT as thermal comfort parameters.

MATERIALS AND METHODS
Two sessions of collecting EEG and PPG data under different thermal conditions were carried out.These took place at the University of Beira Interior during July 2020.Each experimental session had the same five healthy participants.In order to maintain the anonymity, participants were assigned a number: participant 1 was the first to enter a climatized room and have his/her physiological signals collected; participant 2, the second to enter the room, and so on.Their entrance in each of the rooms with the different thermal conditions occurred at a lag.That is, the second volunteer entered 10 minutes after the first, the third 10 minutes after the second volunteer and so on.

Equipments
DeltaOhm Thermal Microclimate HD32.1 equipment and respective probes were used to measure thermal variables.The HD32.1 microclimate measurement unit was designed to analyse the microclimate in the workplace, enabling the detection of parameters required to establish whether a certain location is suitable for carrying out certain activities.In the present study, the operating program used was HD32.1 'A' as it is the most suitable and allows for the measurement of wet bulb temperature.In addition to the direct measurements carried out with the appropriate probes, the equipment also allows for the PMV, PPD and WBGT indexes to be calculated.
To acquire the physiological data, an EEG wearable headband developed at IBEB based on the Bitalino platform was used [26].The headband had two pairs of electrodes in a bipolar montage located at the forehead (Fp1 and Fp2, according to the international 10-20 system), a PPG sensor located at the left earlobe (A1 location) and was connected via Bluetooth to an in-house built mobile application, that recorded and processed the acquired signals [27].

Data acquisition
The two physiological signal collection sessions were carried out under different thermal conditions: one in the morning, in a warmer environment, and the other in the afternoon, in a comfortable environment.Each session lasted about forty minutes for each volunteer, of which the first 30 minutes corresponded to the acclimatization period and the remaining 10 minutes to the acquisition period.To minimize differences between volunteers in terms of mental activity Hunter [20] Yao [21] Carvalho [22] Costa [11] Lan [9] Costa [10] Okamoto [6] Ftaiti [23] Alfa/Beta Shan [13] Choi [24] Davey [25] Alfa/Beta Nybo [12] Minghui [8] *HR: Heart Rate; HRV: Heart Rate Variability; PMV: Predicted Mean Vote; PPD: Predicted Percentage Dissatisfied; WBGT: Wet Bulb Globe Temperature.
during acquisition, they all performed the same task.In this case, the task consisted in performing easy level Sudokus.
To carry out the experimental component of the present study, it was initially necessary to prepare all the material to be used and adjust the environmental variables in each session to obtain the desired environment.According to the room air conditioning control equipment, the variables considered were air temperature and ventilation level (3 levels).Since in session 1 the goal was to achieve a warmer environment, it was enough to turn off the ventilation due to the high outside temperature.In session 2, the purpose was to achieve a different environment from session 1, in this case colder; for that the ventilation level was set at maximum (level 3) and the temperature was set at the minimum value.
Subsequently, the experimental procedure was initiated.First, each volunteer, soon after entering the room, had to remain 30 minutes without collection of physiological signals to acclimatize to the session environment.After this period, the headband was put on and the volunteer was given the task of performing the Sudoku.At the end of the 10-minute period, the headband was removed, and the electrodes were properly disinfected.

Signal Processing
The signal processing was performed by the above-mentioned mobile application [5] in order to obtain the decomposition of the EEG signal in the different frequency ranges as well as to determine the HR and HVR from the PPG signal.The values obtained for each brainwave were computed from the Spectral Power Density (PSD) of the EEG signal, using the Fast Fourier Transform.In particular, the signal power was computed as a function of frequency for each frequency range in one second intervals and averaged over the full duration of the EEG signal.In order to adequately compare the participants' signals, the PSD of each brainwave relative to the whole frequency range was computed as follows Equation (1): 3 RESULTS AND DISCUSSION

Sample characterization
The sample was composed of 5 healthy participants, 2 females and 3 males, aged between 22 and 27 years old.Regarding the level of clothing insulation, it ranged between 0.26 clo and 0.53 clo.As for the activity level of the participants, the same value was adopted for all (1.33 W/m 2 ), since they were all sitting and performing the same task.These last two parameters were automatically calculated using the DeltaOhm software.

Thermal comfort assessment -PMV PPD and WBGT indexes
As previously mentioned, thermal comfort can be assessed by calculating thermal comfort indexes, such as the PMV, PPD and WBGT, which combine one or more environmental variables, such as mean radiant temperature, air temperature, air humidity and air speed.Table 2 shows the thermal comfort indexes of each participant in relation to the two sessions carried out -1 and 2. It was observed that the thermal comfort indexes have different values in sessions 1 and 2 (Wilcoxon test, p<0.05).By analyzing Table 2, it is visible that the WBGT index was equal for all participants in each session, because each session was held in the same room, that is, with the same environmental conditions for all.According to the ISO 7730 standard [28], in thermal comfort conditions, the PMV index is in the range of -0.5 to 0.5.Has can been seen in Table 2, that did not occur in session 1 since the PMV index was greater than 1 for all participants.As for the PPD index, the same standard indicates that an environment is considered thermally comfortable when the percentage of people dissatisfied  with a given environment is less than 10%, which also did not occur during the same session which presented a PPD index greater than 60%.Regarding session 2, four of the five participants showed a PMV value between -0.5 and 0.5, including, whereas the PPD index was lower than 10% for 3 participants.Regarding the remaining participants, volunteer 5 had a PMV index equal to -0.7 and a PPD index slightly higher than 10% for participant 4.
In view of this analysis, it can be concluded that the environment of session 1 was evaluated as a thermally hot environment for all participants since the PMV index was higher than 1.7 and the PPD index was higher than 60%.Although in session 2, two of the participants showed PMV values slightly lower than -0.5 and PPD values slightly higher than 10%, the environment of this session was considered thermally comfortable, since these indexes did not deviate much from the values corresponding to a thermally comfortable environment.

Electrophysiological signals versus thermal comfort indexes
The temporal evolution of the variables studied -relative brainwave PSD of the different frequency ranges, HR and HRV -was first analysed by visual inspection.Figure 1 shows two examples of relative Delta and Beta PSD of participant 1 for the entire record, i.e., for the acclimatization and the acquisition period.In the case of participant 5, as signals showed numerous outliers, it was decided to remove them, i.e., the temporal instant and corresponding values of outliers were eliminated.Thus, in this case, about one minute and forty seconds out of a total of 10 min were removed.
Therefore, the variables acquired during the acquisition period from the physiologic signals of sessions 1 and session 2 -with different thermal conditions and different thermal comfort indexes -could be studied and compared.
Beginning with the analysis of the Delta wave, it is possible to observe in Figure 2 that the relative PSD showed changes between sessions with different thermal comfort indexes.It should be noted that there were statistically significant differences for all participants between session 1 and session 2 (Wilcoxon test, p<0.05) and that the relative PSD of the Delta wave was higher in the session whose environment was evaluated as uncomfortable (session 1).These results are in agreement with the literature [6], [10] which indicates that Delta activity is higher in thermally uncomfortable environments.
In relation to the Beta wave, it is possible to observe in Figure 3 that the relative PSD showed changes between sessions with different thermal comfort indexes.There were statistically significant differences for all participants between session 1 and session 2 (Wilcoxon test, p<0.05) and the relative PSD of the Beta wave was lower in the session whose environment was evaluated as uncomfortable (session 1).
It can also be observed that the relative PSD values were quite similar between participants in each session.In fact, participants were performing a mental task requiring concentration (i.e., sudoku) and the Beta wave is associated with states of alertness and concentration [10].
The Beta differences observed between sessions agree with [10], whose authors concluded that Beta activity tends to be higher in colder environments and lower in warmer environments.However,  this study didn't consider a thermally comfortable environment but two uncomfortable environments -one cold and another hot.According to our results, the colder environment -which is also thermally comfortable -was favorable to higher levels of concentration and attention.
Theta, Alpha and Gamma waves were also studied but they did not show statistically significant differences between sessions 1 and 2.
Regarding cardiac activity, the mean and standard deviation values of HR and HRV are shown in Tables 3 and 4 for all participants and for sessions 1 and 2. Comparing results from session 1 and session 2, HR values were in general higher and HRV values were in general lower in session 1, except for participant 3 (Wilcoxon test.p<0.05).This means that HR reached higher values and HRV lower values in the session with higher thermal comfort indexes (PMV, PPD and WBGT), corresponding to a thermally uncomfortable environment (session 1); whereas, for the session with a thermally comfortable environment (session 2), HR reached lower values and HRV higher values.
On one hand, this tendency agrees with the literature that relate higher HR with thermally uncomfortable conditions [15], [16] and lower HRV with stress situations [18].On another hand, there are some studies that tried to relate HR and HRV with thermal environment, [20] comparing two different scenarios, with high and low temperatures, but not exactly a thermally uncomfortable environment.They concluded that HR increased with temperature.Other authors studied the HRV and concluded that for thermally comfortable environments the HRV was more stable, and for warm environments the HRV had lower values [28] , whilst [21] compared environments with 21ºC, 26ºC and 29ºC and observed that HRV was lower for the mild temperature of 26ºC.
In this way, and although the small size of the sample, our results indicate a possible relation between physiologic response and thermal comfort indexes.

CONCLUSION
This work aimed to study the physiologic response in different thermal comfort environments.Five participants were submitted to different thermal conditions, while electrical brain and cardiac activity were acquired for the analysis of brainwaves from the EEG signal, as well as HR and HRV from the PPG signal.
Although the limitations of our study, like the small sample size, our results seem to indicate the existence of a relation between physiologic response and thermal comfort indexes, in agreement with the literature.Ours results further suggest that in thermally uncomfortable environments Delta activity is higher, Beta activity is lower, as well as HR values were in general higher and HRV values were in general lower for such environments.
This case study will constitute a good working basis for further research with an increased sample size and more variate thermal environments in order to validate the relation between physiologic response and thermal comfort indexes, and to find the more robust physiologic variables to evaluate the thermal comfort.

Figure 1 :
Figure 1: Temporal evolution of relative PSD of Delta and Beta waves, for sessions 1 and 2. The first 30 min correspond to the acclimatization period and the last 10 min (delimited by a grey box) correspond to que acquisition period for which values were studied and compared.

Figure 2 :
Figure 2: Relative PSD of Delta activity and thermal comfort indexes for sessions 1 e 2 and for all participants,

Figure 3 :
Figure 3: Relative PSD of Beta activity and thermal comfort indexes for sessions 1 e 2 and for all participants.

Table 1 :
Research results of the influence of thermal comfort on the amplitude of the physiological signals, used in this work.

Table 2 :
Mean (standard deviation) of thermal comfort indexes of each participant for the two sessions (1 and 2).

Table 3 :
Mean (standard deviation) of HR (bpm) of each volunteer and for sessions 1 e 2.

Table 4 :
Mean (standard deviation) of HRV (ms) of each volunteer and for sessions 1 e 2.