Relationship between human behavior and indoor air quality of private room in a care facility for the elderly in Japan

In Japan, the number of elderly people in need of nursing care is increasing while the population of young people is decreasing, and the potential for labor shortages in the field of elder care is of great concern. This study aimed to estimate the behavior of the elderly by using sensors to monitor indoor air quality (IAQ), without placing undue burden on the elderly or their caregivers. Odor and carbon dioxide (CO2) concentrations were monitored in a private room of a nursing home in the Kanto Region of Japan, the behaviors of the resident and staff members were recorded, and the relationship between the two was analyzed. Both odor and CO2 concentrations were higher when the resident was present than when absent, indicating that the resident was one of the main sources of indoor odor and CO2. In addition, after the resident entered the room, the CO2 concentration increased and remained stable, whereas the odor concentration tended to vary after the resident entered the room, first increasing and later decreasing. This suggested that the increase or decrease in odor could be used to monitor the behavior of the resident and staff members. The relationship between the slopes of odor and CO2 in typical behavioral events suggest that if only odor increases and CO2 does not change, the likelihood of the event in which feces were observed during diaper changes is high. In addition, based on the behavior near the sensor, the rate of CO2 and odor emissions differed between the elderly resident and the younger staff members, suggesting that the ratio of odor slope to CO2 slope may be greater in the elderly than in younger people. Furthermore, the repeated number of increases and decreases in odor and CO2 suggested that multiple events could be distinguished. These results suggest that IAQ can be utilized to estimate the behavior of residents and staff in nursing care facilities for the elderly.


Introduction
As of October 1, 2021, Japan's total population stood at 125.5 million (Cabinet Office Japan, 2022a), while the population of older adults aged ≥ 65 years was 36.21 million, accounting for 28.9% of the total population (Cabinet Office Japan, 2022a).By 2036, it is estimated that one out of every three people in Japan will be ≥ 65 years of age, as the number of older adults increases while the total population declines (Cabinet Office Japan, 2022a).
The number of people certified as requiring nursing care or support under Japan's long-term care insurance system was about 6.6 million in FY 2019, an increase of about 1.9 million from FY 2009(Cabinet Office Japan, 2022b).Because this trend is expected to continue, the demand for elder care workers is also expected to increase in the future.However, the working-age population in Japan is expected to decrease as the total population decreases, and the potential for labor shortages in the field of elder care is of great concern.
One approach to compensate for such labor shortages is the application of various technologies.For example, socially assistive robots (SARs) have been developed for the purpose of caring for the elderly, and their application and potential issues have been discussed (e.g., Vandemeulebroucke et al., 2021).In addition, nursing equipment has been developed that can accurately identify and deal with urine and feces (Hu et al., 2023).
In recent years, there has also been a growing trend toward the use of the Internet of Things (IoT) to provide for the welfare of the elderly.IoT devices such as Wi-Ficonnected sensors are capable of automatically collecting and transmitting data in real-time to servers with minimal or no human intervention (Gupta, 2016;Perez et al., 2023).For example, fall-detection systems (Cardenas et al., 2023), fall-prediction systems (Kulurkar et al., 2023), and gait episode identification systems (Peimankar et al., 2023) based on artificial intelligence, deep learning, and wearable sensors have been developed in order to prevent falls in the elderly.Nakajima et al. (2014) also developed a system that informs the user of the presence or absence of excrement by attaching a reusable sensor to a diaper.Other studies have reported methods for confirming the presence or absence of feces by attaching sensors to diapers (Baek, 2023;Yang et al., 2008;Simik et al., 2014).The use of such wearable sensors is significantly less expensive than large-scale equipment such as robots, and thus has the potential to improve the quality of life in the elderly while reducing labor costs.
However, wearing a sensor may be uncomfortable for some elderly people, and the sensor might become detached due to carelessness or other reasons.In addition, when sensors are attached to clothing or diapers, they must be reattached each time they are changed, which can be time-consuming.
Efforts to understand the behavior of nursing home residents based on changes in indoor air quality (IAQ) have also been reported (e.g., Deng et al, 2021;Hattori et al., 2022;Madokoro et al., 2021;Reis et al., 2023;Sawada et al., 2001;Tanaka & Munaka, 2021).For example, Hattori et al. (2022) found that changes in data collected by a CO 2 sensor installed indoors and a temperature sensor installed outdoors could be used to determine the indoor heating status.Tanaka and Munaka (2021) found that volatile organic compound (VOC) concentrations in private rooms of nursing homes increased before diapers were changed due to excretion.They reported that VOC concentrations could be used as an indicator of defecation behavior in the elderly.Thus, because changes in IAQ are thought to be closely related to certain behaviors in the elderly, IAQ data could be utilized to ascertain the behavior of elderly residents without placing undue burden on them or their caregivers.
Sensors for monitoring IAQ include CO 2 , relative humidity, temperature, VOC, and odor sensors.It has been reported that data collected by CO 2 , relative humidity, and temperature sensors can be utilized to understand human behavior (e.g., Deng et al., 2021;Hattori et al., 2022;Madokoro et al., 2021;Reis et al., 2023).In contrast, there are very few reports on the relationship of VOC and odor sensors with human behavior.However, VOCs and odors are mixtures of various organic and inorganic compounds, and changes in their concentrations are thought to contain a great deal of information on the behavior and health of residents in a given room.A careful analysis of changes in VOC and odor in the room may be predictive of changes in human behavior and health.
Therefore, the purpose of this study was to clarify the effects of certain behaviors on indoor air quality by simultaneously monitoring IAQ and the behavior of a resident and staff members in a private room in a nursing home.IAQ was monitored using CO 2 and odor sensors.

Ethical considerations
Written informed consent was obtained from the resident and the management of the nursing home prior to their participation.In addition, this study was conducted in accordance with the Ethical Guidelines for Research Involving Human Subjects set forth by the Central Research Institute of Electric Power Industry, to which the corresponding author belongs.

Monitoring site
The indoor air was monitored in a private room of a three-story reinforced concrete nursing home located in the Kanto Region of Japan, the details of which were described previously (Tanaka & Munaka, 2021).The layout of the participant's room is shown in Fig. 1 and Table 1.The resident needs assistance to stand up and sit down.The resident always wears a diaper and does not use the toilet in his private room.The resident eats all three meals in the communal dining room and takes a bath once every three days.Except for having meals and bathing, the resident spends most of his time lying in bed.After returning from a meal, the resident usually brushes his teeth at location A in Fig. 1 and then watches TV at location B in Fig. 1 while sitting in his wheelchair.

Sample and data collection
Sample and data collection was performed on 30 consecutive days.A CO 2 sensor (2JCIE-BU01, OMRON) and an odor sensor (TGS2603, Figaro) were placed on a shelf at a height of about 1 m, and data were recorded every 10 s (Fig. 1).Both were oxide-semiconductor sensors.The odor sensor had a high sensitivity to amines and sulfur compounds but a low sensitivity to ammonia.Two infrared cameras were installed on the wall (Fig. 1) at a height of about 1.8 m to capture images of the room every 10 s in order to monitor the behavior of the resident and staff members, the opening and closing of the sash window and the door, and the use of the air conditioner.

Overview of indoor air quality
A summary of the IAQ for the entire monitoring period is shown in Table 2.The average (standard deviation) for the entire monitoring period was 877 (± 105) ppm for CO 2 and 41.0 (± 29.1) for odor.CO 2 and odor were Fig. 1 Layout of the resident's private room in the nursing home.After returning from a meal, the resident often brushed his teeth at location A and then watched TV while sitting in his wheelchair at location B Table 1 Characteristics of the private room and the resident 1.15 and 1.13 times higher, respectively, when the resident was present compared with when the resident was absent.The results of this study are consistent with previous reports on the main sources of CO 2 as well as VOCs, which are one of the causes of odors.The percentage of CO 2 exceeding 1000 ppm was 7.8% for the entire period (Table 2).In Japan, the IAQ standard for CO 2 is less than 1000 ppm (Ministry of Health, Labour & Welfare, 2023).Indoor CO 2 concentrations can quickly rise due to an increase in the number of people in the room, use of heating, and inadequate ventilation, and it is not uncommon for CO 2 concentrations to exceed 1000 ppm.In the winter in particular, there have been reports of cases exceeding 5000 ppm when combustion heating is used (Hattori et al., 2022).In the nursing home examined in the present study, combustion heating was not used either in the private rooms or in other spaces, and thus no combustion-derived CO 2 was generated.In addition, the two ventilation openings in the private rooms (see Fig. 1) would be expected to provide adequate ventilation, thereby keeping the CO 2 concentration low in the room.

Relationship between changes in CO 2 and odor over time and behavior of the resident
Examples of IAQ changes over time and the behavior of the resident and staff members are shown in Fig. 2 (see Additional file 1: Figure S1 for full data).Both CO 2 and odor tended to increase rapidly immediately after the resident or staff members entered the room and to decrease when the resident or staff members left.In other words, CO 2 and odor often tended to increase or decrease at the same times each day.This further supports the observation mentioned in Sect.3.1 that one of the main sources of CO 2 and odor in the private room was the resident.However, while CO 2 remained the same after suddenly increasing after the resident entered the room, odor tended to increase and decrease repeatedly after increasing when the resident entered the room (e.g., from 0:00 to 4:00 and 6:00 to 10:00 on 2/18, from 0:00 to 4:00 and 19:00 to 21:00 on 2/24, from 19:00 to 21:00 on 2/25, and from 1:00 to 5:00 on 3/3).These results suggest that odors may be capturing some changes or events in the room, including the resident's behavior, unlike CO 2 .
In the following sub-sections, we discuss the relationships between changes in CO 2 and odor concentrations and several characteristic behaviors of the resident.

Feces
Figure 3 shows the time variation in IAQ and the behavior of the resident and staff members before and after times when feces were observed during diaper changes (denoted as F in the figure).During the observation period, feces were observed 17 times during diaper changes in the resident's private room.The patterns of IAQ changes before the F event are shown in Table 3.The most frequently observed pattern before the F event was one in which an increase in odor was detected and continued to rise while CO 2 remained nearly the same (Pattern 1), accounting for 10 of the total 17 observations.The next most frequently observed pattern was one in which the odor rose before the F event and then decreased but remained at a higher value than before the increase began while CO 2 remained nearly the same (Pattern 2), which occurred in 3 of the 17 observations.There was also a pattern (Pattern 3: n = 1) in which the odor increased gradually with repeated increases and decreases while CO 2 remained nearly the same.Additionally, there was a pattern (Pattern 4: n = 2) in which the odor increased and then began to decrease, becoming lower than at the start of the increase, while CO 2 remained nearly the same.These four patterns were similar in that odor increased before the F event and CO 2 remained the same, and they accounted for 16 of the 17 observations (94%).In contrast, another pattern was that both odor and CO 2 remained nearly the same (Pattern 5), which occurred in only 1 of the 17 observations.In Pattern 5, the odor concentration rose sharply when the  .This odor may have reduced the influence of the odor originating from the subsequent excretion of feces, and the resulting increase in odor may not have been detected.These results revealed that odor tended to increase before the F event, whereas CO 2 did not.The trend observed here was considered reasonable because the excretion of feces would naturally contribute to an increase in odor in the room, whereas CO 2 emissions due to excretion would be negligible.Thus, the results demonstrate that detection of an increase in odor concentration can be utilized to infer that an excretion of feces has occurred.
It has previously been reported that sensors can detect feces.For example, Oyabu et al. (2001) reported that odor sensors installed in a lavatory can detect feces and urine, and that the rate of change depends on the sensor and the subject.This suggests that odor sensors can infer excretion of feces.The potential for estimating behavior based on the rate of change in odor and CO 2 is discussed in Sect.3.3.

Urine
Figure 4 shows examples of IAQ variation and the behavior of the resident and staff members before and after times when only urine was observed during diaper changes (denoted as U in the figure).During the observation period, there were 158 U events, and the variation in IAQ before the U event was broadly classified into the following five patterns, with the numbers in parentheses indicating the number of events corresponding to each pattern.Pattern 1 (Fig. 4a): both odor and CO 2 remained nearly the same (n = 56); Pattern 2 (Fig. 4b): both odor and CO 2 increased (n = 21); Pattern 3 (Fig. 4c): odor increased while CO 2 remained nearly the same, even after diaper changes (n = 22); Pattern 4 (Fig. 4d): odor decreased while CO 2 remained nearly the same (n = 22); Pattern 5 (Fig. 4e): odor increased while CO 2 remained nearly the same and then odor decreased after diaper changes (n = 22).There were 143 events that fell into these five patterns, accounting for more than 90% of the total.In addition, there was a pattern in which both odor and CO 2 decreased (n = 9) and another pattern in which odor decreased while CO 2 increased (n = 5).If, like feces, the excretion of urine affects IAQ, then it is expected that odor alone would increase after the U event and then decrease after diaper changes.Odor alone increased before diaper changes in Patterns 3 and 5, but in Pattern 3, odor also increased after diaper changes, suggesting that the effect of odor caused by urine was small.This suggests that Pattern 5 is the one in which the odor sensor may capture the increase in odor caused by urine; however, only 22 events corresponded to this pattern, accounting for only 14% of the total.Pattern 1 was the most common, accounting for 35% of the total.Taken together, these results indicate that it may be difficult to estimate the timing of urine excretion based solely on the sensor data collected in this study.The reason why the sensor did not capture the odor originating from urine may be that urine is absorbed by the polymer of the diaper, making it difficult for the odor to leak out of the diaper.As mentioned earlier, the odor sensor used in this study has a low sensitivity to ammonia, which is one of the main odor-causing substances in urine; this low sensitivity may also be a factor in the inability of the sensor to clearly capture F events.
However, urine leakage is considered to be accompanied by a strong odor.For example, the main odors reported for people with urinary incontinence living in elderly care facilities are 3-methylbutanal, trimethylamine, cresol, guaiacol, 4,5-dimethylthiazole-S-oxide, diacetyl, dimethyl trisulfide, and 5-methylthio-4-penten-2-ol (Hall et al., 2017).The odor sensor used in the present study is highly sensitive to amines and sulfur compounds, and is thus expected to detect urine leakage.

Resident's proximity to the sensor
Figure 5 shows examples of IAQ variation when the resident was sitting in a wheelchair and watching TV near the sensor (denoted as S in the figure).During the measurement period, S events were observed 52 times.Both odor and CO 2 usually increased during S events, with odor being particularly noticeable (Pattern 1: Fig. 5a).There were 43 S events in which both odor and CO 2 increased, accounting for 83% of all S events.Note that both odor and CO 2 increased for a certain period and then decreased, which coincided with the timing of the resident's transfer from the wheelchair to the bed.Because the bed is somewhat far from the sensor (see Fig. 1), it is reasonable that odor and CO 2 would decrease when the resident moves to the bed.However, odor and CO 2 did not always increase simultaneously during an S event.In the case shown in Fig. 5b (Pattern 2), CO 2 increased during the S event but odor did not.In another case, odor increased during the S event while CO 2 decreased (Fig. 5c: Pattern 3).In these two cases, odor and CO 2 fluctuated significantly before the resident entered the room, due to ventilation from opening windows (denoted S in Fig. 5b) and staff members working in the room (Fig. 5c), the details of which are described below.
In Pattern 1, the average rate of increase in odor (-/s) ranged from 0.0025 to 0.054 and differed significantly from event to event.In general, possible sources of odor originating from humans include exhaled breath and body odor.Of these, exhaled odors are thought to vary with diet.In other words, the odor from exhaled breath is expected to be stronger after eating food with strong aromatic ingredients such as garlic (Mirondo & Barringer, 2016;Sato et al., 2020).All S events were timed immediately after the residents returned to their private rooms from their meals, and the intensity of odor in the exhaled breath was considered to be strongly influenced by the preceding meal.In addition, body odor may be related to the time after bathing.The resident bathed once every 3 days in the nursing home.
Table 3 Observed patterns of changes in IAQ preceding feces events

Staff members entering and working in the room
Figure 6 shows examples of IAQ variation when staff members entered the private room (indicated by the light blue line in the figure).The patterns of IAQ change when staff members entered the room include cases where both odor and CO 2 increased (Pattern 1: Fig. 6a, b), cases where neither odor nor CO 2 changed (Pattern 2: Fig. 6c), and cases where either odor or CO 2 decreased or they both decreased (Pattern 3: Fig. 6d).
Staff members entered the room 530 times during the observation period: 499 times when the resident was present and 31 times when the resident was not.When the  3) other (e.g., collecting laundry, making rounds, administering medication).When the resident was not present, the staff members performed two main types of work: (4) cleaning and ( 5) other (e.g., collecting laundry).Of these tasks, (1), (2), and (4) involved strong physical activity and could take about 10 min or more to complete.The changes in IAQ at these times often followed Pattern 1 (Fig. 6a, b).In contrast, collecting laundry and administering medication involved light physical activity and often took only a few minutes.In these cases, IAQ movements were classified mostly as Pattern 2 (Fig. 6c) or Pattern 3 (Fig. 6d).The IAQ variations shown in Fig. 6a, b are thought to reflect such physical activity.In other words, the results indicate that work involving strong physical activity can be estimated from changes in IAQ.

Heavy movement of the resident in sleeping
Figure 7 shows examples of IAQ variation when the resident moved his arms and legs, for example while removing part of the bedding while sleeping (denoted as H in the figure: n = 67) and when the resident was sleeping peacefully (denoted as P in the figure).When such physical activity was observed while the resident was in bed, there were two patterns of IAQ variation: one in which odor increased and decreased for a short Fig. 5 Examples of changes in indoor air quality when the resident was sitting in a wheelchair and watching TV near the sensor.S: The resident was sitting in a wheelchair and watching TV near the sensor.V: Staff members opened the windows time while CO 2 did not change significantly (Pattern 1: Fig. 7a; n = 50), and another in which odor increased monotonically and CO 2 remained nearly the same (Pattern 2: Fig. 7b; n = 16).Pattern 1 accounted for 75% of the total.Note that in a very small number of cases, odor was reduced (n = 2).Intense physical activity is thought to promote sweating, and it was also considered that removing bedding would facilitate the diffusion of body odors into the indoor air.The image analysis also revealed that the resident's physical movements were often intermittent.The repeated increase and decrease in odor over a short period of time was considered to be due to the abovementioned reasons.However, in cases where odor increased monotonically, the excretion of feces was often confirmed during subsequent diaper changes.Because feces are generally considered to have a stronger odor compared with body odor or exhaled breath, the odor would continue to increase if the odor was diffused into the indoor air such as by removing the bedding.The fluctuation of odor and CO 2 during sleep has also been reported by Nagamune et al. (2021)  Meanwhile, IAQ when the resident was sleeping peacefully (n = 50) was dominated by cases in which the odor was nearly the same or decreased while CO 2 was nearly the same (Fig. 7c), accounting for 74% of the total.This Taken together, these results suggest that the degree of physical activity at bedtime can be estimated from fluctuations in IAQ.

Ventilation
Figure 8 shows examples of IAQ variation when staff members opened the windows.The target facility is airconditioned, and each room is equipped with natural ventilation openings and ventilation fans that are always in operation, so the private room is ventilated even without opening the windows.However, windows may be opened during diaper changes involving feces or during cleaning to more quickly change the air.During these events, both odor and CO 2 decreased rapidly (Fig. 8).
There have been many reports on the rapid decrease in indoor CO 2 due to ventilation (e.g., Emmerich & Persily, 1997), and the results of this study are consistent with those previous findings.In addition, the rapid decrease in odor and CO 2 during ventilation due to window opening was not observed during other events, suggesting that it is easy to infer when windows were opened based on the changes in IAQ.

Potential for estimating residents' and staff members' behavior based on changes in indoor air quality
Figure 9 shows the relationship between the slope of odor and that of CO 2 during typical events; the three events with positive rates of change in odor are F, S, and T. Thus, a continuous increase in odor is likely to be one of these three events, and if CO 2 does not increase, it is likely to be an F event.In contrast, an S event, in which the resident is the source, and a T event, in which the younger staff member is the source, are relatively close together,  but is more likely to be S if the odor slope is greater.This may be because the resident in this study is older and older people tend to have stronger halitosis and body odor (e.g., Egusa, 2000;Mitro et al., 2012); in addition, the resident is more prone to odor production because they were wearing diapers (e.g., Tokoro et al., 2015).These results suggest that there is a difference in CO 2 and odor emission rates between older and younger adults, and that the ratio of odor slope to CO 2 slope (odor slope / CO 2 slope) may be greater for older adults than for younger adults.Meanwhile, halitosis is most often a consequence of oral bacterial activity, typically from anaerobes (Scully & Rosenberg, 2003).This can sometimes be caused by systemic disease, and some patients may have a psychogenic background (Scully & Rosenberg, 2003).Also, as previously mentioned, oral odor is stronger after eating aromatic foods such as garlic (Mirondo & Barringer, 2016;Sato et al., 2020).In other words, it is important to note that the intensity of halitosis depends not only on age, but also disease and diet.Figure 9 also shows that U, H, and P events had a lower percent change in odor.Among these events, U had a mean rate of change (ppm s −1 ) of 0.13 for CO 2 , while H and P had a mean rate of change of − 0.01 and 0.00, respectively, but the standard deviation of the CO 2 slope in the U event was ± 1.35, which is extremely large compared with the other events.Therefore, these three events are difficult to distinguish from the slopes of CO 2 and odor alone.Figure 10 shows the results of counting the number of positive and negative changes (n of P/N change) in the slope of odor as well as that of CO 2 for these three events, focusing on the fact that both CO 2 and odor increased and decreased repeatedly in the H event. Here, when the same sign continued for more than 10 s it was judged that the positive and negative slopes had changed.The figure shows that in the H event, the slopes of both odor and CO 2 repeatedly increased and decreased during the event, whereas in the P event, CO 2 repeatedly increased and decreased, while odor did not change much.In addition, the U event showed the smallest increase and decrease in CO 2 among the three events.These results suggest that U, H, and P events can be distinguished by considering the number of positive and negative changes in the slopes of CO 2 and odor, in addition to the slopes themselves.
The slopes of odor and CO 2 at each event are expected to depend on a variety of factors, including the size of the room, ventilation conditions, occupant health, physical activity level, presence of roommates, and diet.In particular, halitosis and body odor can vary widely depending on the individual and sex.For example, the "uremic breath" of patients with end-stage renal disease contains high levels of dimethylamine and trimethylamine (e.g., Simenhoff et al., 1977), which give off a strong odor.In addition, contracting an infectious or metabolic disease often results in a change in body odor (e.g., Shirasu & Touhara, 2011).A study investigating 373 VOCs in human axillary odor have also found individual and sex-specific VOCs (Penn et al., 2007).Therefore, it should be noted that the slope of odor and CO 2 for each event presented in this study may vary depending on the situation.

Conclusion
The purpose of this study was to clarify the impact of human behavior on IAQ by observing both IAQ and human behavior in a private room of a nursing home.Both odor and CO 2 concentrations were higher in the room when the resident was present than when the resident was absent, indicating that the resident was one of the main sources of indoor odor and CO 2 .Both odor and CO 2 concentrations rose sharply when the resident entered the room and decreased when the resident left.However, after the resident entered, CO 2 increased and then remained nearly unchanged, whereas odor tended to increase and decrease frequently, suggesting that the behavior of the resident and staff members affected IAQ.
The relationship between the slopes of odor and CO 2 in typical behavioral events suggest that if only odor increases and CO 2 does not change, the likelihood of an F event is high.In addition, based on the behavior near the sensor, the rate of CO 2 and odor emissions differed between the elderly resident and the younger staff members, suggesting that the ratio of odor slope to CO 2 slope (odor slope/CO 2 slope) may be greater in the elderly than in younger people.Furthermore, the results suggest that it may be possible to use the number of repetitions in increase/decrease as an index for distinguishing between U, H, and P events, which are difficult to distinguish only by the rate of change in IAQ.Taken together, the results of this study suggest that certain behaviors such as defecation as well as the kind of work performed by staff members can be estimated from variations in IAQ.

Fig. 2
Fig. 2 Examples of time variation in indoor air quality and the behavior of the resident and staff members in the resident's room.F: Feces were observed during diaper changes.S: The resident was sitting in a wheelchair and watching TV near the sensor.U: Only urine was observed during diaper changes.V: Staff members opened the windows

Fig. 3
Fig. 3 Time variation in indoor air quality when feces were observed during diaper changes.F: Feces were observed during diaper changes.S: The resident was sitting in a wheelchair and watching TV near the sensor.U: Only urine was observed during diaper changes.V: Staff members opened the windows

Fig. 4
Fig. 4 Examples of changes in indoor air quality when only urine was observed during diaper changes.S: The resident was sitting in a wheelchair and watching TV near the sensor.U: Only urine was observed during diaper changes.V: Staff members opened the windows and Yagi et al. (2021).Nagamune et al. examined the behavior of participants in a model house and monitored the fluctuation in CO 2 and VOCs detected by sensors, and suggested the possibility of alcohol consumption before sleep as a reason for the increase or decrease in VOCs during sleep.The participants in Nagamune et al. 's experiment were young to middle-aged healthy individuals, and their lifestyles were considered different from those of the elderly participants in the present study.Therefore, the cause of the variation in odor and CO 2 during sleep in the present study may be highly dependent on the lifestyle of the resident.

Fig. 6
Fig. 6 Examples of changes in indoor air quality when staff members entered the resident's room.C: Staff members cleaned the room.T: Staff members transferred the resident.U: Only urine was observed during diaper changes

Fig. 7
Fig. 7 Examples of changes in indoor air quality when the resident was sleeping in his bed.F: Feces were observed during diaper changes.H: The resident moved his arms and legs.P: The resident slept peacefully

Fig. 8
Fig. 8 Example of changes in indoor air quality when staff members opened the windows.F: Feces were observed during diaper changes.V: Staff members opened the windows

Fig. 9
Fig. 9 Relationship between slope of odor and that of CO 2 during typical events.Error bar: Standard deviation.F: Feces were observed during diaper changes.S: The resident was sitting in a wheelchair and watching TV near the sensor.V: Staff members opened the windows.H: The resident moved his arms and legs.U: Only urine was observed during diaper changes.T: Staff members transferred the resident.P: The resident slept peacefully

Fig. 10
Fig. 10 Relationship between the number of positive and negative changes (n of P/N change) in the slope of odor and that of CO 2 for H, P, and U events.Error bar: Standard deviation.H: The resident moved his arms and legs.P: The resident slept peacefully.U: Only urine was observed during diaper changes

Table 2
Summary of IAQ for the entire monitoring period observed Because the sensitivity of odor sensors varies depending on the substance, it cannot be converted to a concentration when several substances are mixed together; therefore, the values are expressed here as dimensionless signal valuesTanaka et al.Asian Journal of Atmospheric Environment  (2023) 17:11 a Standard deviation b Ratio of concentration when resident was present (P) to concentration when resident was absent (A) c