Determination of Waste Management Workers’ Physical and Psychological Load: A Cross-Sectional Study Using Biometric Data

Waste management workers experience high stress and physical strain in their work environment, but very little empirical evidence supports effective health management practices for waste management workers. Hence, this study investigated the effects of worker characteristics and biometric indices on workers’ physical and psychological loads during waste-handling operations. A biometric measurement system was installed in an industrial waste management facility in Japan to understand the actual working conditions of 29 workers in the facility. It comprised sensing wear for data collection and biometric sensors to measure heart rate (HR) and physical activity (PA) based on electrocardiogram signals. Multiple regression analysis was performed to evaluate significant relationships between the parameters. Although stress level is indicated by the ratio of low frequency (LF) to high frequency (HF) or high LF power in HR, the results showed that compared with workers who did not handle waste, those who did had lower PA and body surface temperature, higher stress, and lower HR variability parameters associated with higher psychological load. There were no significant differences in HR, heart rate interval (RRI), and workload. The psychological load of workers dealing directly with waste was high, regardless of their PA, whereas others had a low psychological load even with high PA. These findings suggest the need to promote sustainable work relationships and a quantitative understanding of harsh working conditions to improve work quality and reduce health hazards.


Introduction
Waste management workers responsible for collecting, transporting, and sorting waste face various challenges [1,2]. They face numerous occupational risks, including long working hours, exposure to physical, chemical, mechanical, biological, ergonomic, and social risks, and frequent occupational accidents, which result in physical or mental illness [3]. The United States Department of Labor Occupational Safety and Health Administration includes toxic metals, crushing hazards, and dangerous energy release, among the potential hazards of working with waste [4]. In 2020, the frequency (number of fatalities and injuries due to occupational accidents per one million hours of total actual work) and intensity rates (number of lost workdays per one million hours of total actual work, i.e., the severity of the network using wireless access points (WiFi and 4G-based transfer devices) installed in the work area.
A flowchart of the background, objectives, and analysis in this study is presented in Figure 2, and the study was conducted according to this flowchart.  Preparation for analysis: -Apply t-tests for normally distributed measurement parameters and Mann-Whitney U-tests for non-normally distributed measurement parameters.
-Confirm measurement parameters for which there is a statistically significant difference between workers who handle waste and those who do not.

Background:
By applying the biometric measurement system to workers at waste treatment facility, the working conditions that are expected to cause severe psychological and physical load will be assessed, and the health and safety management of waste workers will be discussed.
Objectives: -Collect and analyze biometric data to understand the differences in working conditions between workers who handle waste and those who do not.
-To confirm the characteristics that appear in the biometric information and characteristics of workers, and to improve the work environment by quantifying the working conditions. Analysis approach 1: Analyze the effect on workers' body surface temperature.
-Multiple regression analysis is used to evaluate the independent variables that are statistically significant among the measured parameters for the dependent variable (body surface temperature).
-Single regression analysis is used to evaluate the independent variables that are statistically significant for the dependent variable (body surface temperature).

Analysis approach 2:
Analyze effects on physical and psychological load. -Evaluate statistically significant independent variables among the measurement parameters for the dependent variables (physical and psychological load).
-Multiple regression analysis is used to evaluate the significant measurement parameters for physical and psychological load and to obtain multiple regression equations. network using wireless access points (WiFi and 4G-based transfer devices) installed in the work area.
A flowchart of the background, objectives, and analysis in this study is presented in Figure 2, and the study was conducted according to this flowchart.  Preparation for analysis: -Apply t-tests for normally distributed measurement parameters and Mann-Whitney U-tests for non-normally distributed measurement parameters.
-Confirm measurement parameters for which there is a statistically significant difference between workers who handle waste and those who do not.

Background:
By applying the biometric measurement system to workers at waste treatment facility, the working conditions that are expected to cause severe psychological and physical load will be assessed, and the health and safety management of waste workers will be discussed.
Objectives: -Collect and analyze biometric data to understand the differences in working conditions between workers who handle waste and those who do not.
-To confirm the characteristics that appear in the biometric information and characteristics of workers, and to improve the work environment by quantifying the working conditions. Analysis approach 1: Analyze the effect on workers' body surface temperature.
-Multiple regression analysis is used to evaluate the independent variables that are statistically significant among the measured parameters for the dependent variable (body surface temperature).
-Single regression analysis is used to evaluate the independent variables that are statistically significant for the dependent variable (body surface temperature).

Analysis approach 2:
Analyze effects on physical and psychological load. -Evaluate statistically significant independent variables among the measurement parameters for the dependent variables (physical and psychological load). -Multiple regression analysis is used to evaluate the significant measurement parameters for physical and psychological load and to obtain multiple regression equations.

Participants
The data were collected from 3 to 4 September 2019, and 24 to 26 August 2020, at the Kyoto Environmental Conservation Corporation, a waste management facility in Fushimi-Ku, Kyoto, Japan. The participants were recruited from among the facility's employees; workers who responded to the call from the facility took part in this experiment. The participants were defined as healthy adults aged 20-63; those with neurological or cardiovascular diseases were excluded. All the participants were men. A total of 29 workers (average age: 35, standard deviation (SD): 15.7 years; average body mass index (BMI): 21.5, SD 3.2; average work experience (EXP): 7.7, SD: 4.2 years) were included in the final sample.
Based on their job profile or responsibilities, the participants were allocated to two groups. Group A (n = 22) was responsible for the transport, dismantling, and sorting of industrial waste, and Group B (n = 7), for the maintenance and repair of the facility. The list of participants in each group is presented in Appendix C. Group A workers were responsible for unloading waste from trucks at the facility, manually arranging it, and then moving it by cart to the loading dock for incineration. In the waste sorting process, relatively small waste articles (not suitable for heavy machinery) were sorted manually, thus posing a continuous exposure risk. Group B workers were tasked with inspecting and repairing equipment at the facility and did not come into direct contact with waste. The difference in the number of Group A and Group B workers is due to the staffing of the facility where this experiment was conducted.

Protocols
Before data collection, the research team tested the sensing wear. They confirmed that it was a non-invasive measurement tool that posed no risk to the wearer, such as interference or discomfort with daily work tasks. An informed consent form was distributed to all the participants prior to data collection. It included an explanation of their rights and assured them of the confidentiality of their data. To minimise the risk of personal information leaks, personal identification codes (identifiers) were assigned to each participant during the experiment and data analysis.

Analysis Tools and Statistical Tests
Statistical analysis was conducted using SPSS Version 26 for Windows (IBM Corp., Armonk, NY, USA) and Excel add-in software Bell Curve (Social Survey Research Information Co., Ltd., Tokyo, Japan) for Excel version 3.21 (Microsoft Corporation, Redmond, WA, USA). In addition, the following statistical tests were performed on the three hypotheticals: Hypothesis 4 (H4): When comparing waste treatment workers to non-treatment workers, we tested whether there were significant differences in the biological parameters of the two populations of workers. The normality of data was tested using the Shapiro-Wilk and Kolmogorov-Smirnov tests (Appendix G). First, the normality between the data in the two populations was tested. If there was normality, a T-test was performed; if not, a Mann-Whitney U test was performed.

Hypothesis 5 (H5):
To analyse the relationship between changes in biological parameters and temperature fluctuations that cause heat stroke and other physical ailments experienced by waste treatment workers, the effect of workers' biological parameters on body surface temperature was analysed. Multiple regression analysis was performed, with the independent variable being the biological parameters of the workers and the dependent variable being body surface (BS) TEMP.

Hypothesis 6 (H6):
To investigate whether the measured biological parameters of waste treatment workers could be used to estimate workload and psychological load, a multiple regression analysis was conducted with the workers' biological parameters as independent variables and physical and psychological load as dependent variables. The multiple regression analysis conducted in this study investigated the presence or absence of multicollinearity among candidate variables and proceeded by combining dependent and independent variables with no multicollinearity.

Body Temperature and HR
Human performance is affected by various environmental factors in a working system, including heat stress [18,19]. The measurement of body temperature is an effective way to observe heat stress in workers in hot environments. However, invasive methods such as rectal and oesophageal measurements are often impractical [20,21]. Humans need to dissipate excess heat generated by the body to maintain thermal equilibrium with their internal temperature maintained at approximately 37 • C. However, the uniforms provided by waste management companies often do not provide adequate protection, especially in the summer when workers are thinly clad. Thus, metabolic heat from the body, environmental factors (temperature, humidity, radiant heat, etc.), and the overall heat load from clothing can result in heat stress [22]. When the core body temperature increases due to poor heat dissipation, the physiological burden of heat can cause health problems such as heatstroke, heat exhaustion, and heat cramps [23]. The physiological process for maintaining body temperature becomes inadequate when personal risk factors are added to excessive environmental or metabolic heat stress, thus resulting in an elevated body temperature and pulse rate and weight loss due to dehydration [24]. If early signs of heat-related illness are disregarded, heat stress can lead to poor work performance and also heat-related injuries and death [25].
Studies using non-invasive wearable devices have found a significant correlation between human HR and body surface temperature measurements and the thermal environment [26][27][28][29]. Eggenberger et al. [30] investigated different exercises and clothing conditions in a hot and humid environment and found two measured parameters, HR and body surface temperature at the scapular region, suitable for predicting rectal temperature. The air layer trapped inside a garment creates a specific microclimate around the body [31], influencing perceived comfort [32]. The microclimate inside a garment influences human satisfaction and performance during activities [33]. Thus, HR and ambient temperature in clothing may be used as indicators of human thermal sensation and heat stress. However, few studies have investigated the effects of HR and ambient temperature on the heat stress of workers.

HRV Metrics
Heart rate variability (HRV) is not only related to its effect on physical load, but also to autonomic control, such as self-regulation and psychological and physiological stress, making it useful for psychological load analysis [14,34]. A low HRV index indicates inadequate coordination between the sympathetic and parasympathetic nervous systems and is a reliable predictor of future cardiovascular diseases [35,36]. Therefore, HRV analysis provides important information for the assessment of physical function and helps identify the risk of physical fatigue and debilitation [37]. The time-domain index of HRV quantifies the extent of variation in the heart rate interval (RRI), which is the time between successive heartbeats, and the frequency-domain index can be obtained from the power spectrum density of a specific frequency band of RRI data. HRV time-domain indicators include the standard deviation of the normal heart rate interval time (NN), called SDNN, the root mean square successive difference (RMSSD), the index of difference between adjacent normal heart rate intervals greater than 50 ms (NN50), and the index of the percentage of the difference between adjacent normal heart rate intervals greater than 50 ms (pNN50). The main frequency indicators are low frequency (LF) power and high frequency (HF) power [14].
In a previous study, RMSSD was related to workers' perception of mental stress [38], with lower values indicating higher stress. The RMSSD metric is not significantly affected by the number of missing data points, thus indicating its robustness for assessing patients with poor data quality. The standard deviation of the RRI (SDRR) is calculated from the SD of the normal RR interval; a lower SDRR indicates lower HRV [39]. Taelman et al. [40], who explored the interaction between HRV and mental stress, reported significantly lower NN50 and mean RR intervals in mentally demanding tasks.
HRV metrics have shown promise in multiple applications for healthcare professionals, and several researchers have focused on this field [41][42][43][44]. Despite concerns about the validity of some HRV indices for measuring sympathetic and parasympathetic nerve activity [45,46], the significance of HRV analysis is supported by many studies, such as LF (0.04-0.15 Hz) power, HF (0.15-0.40 Hz) power, and their ratios. LF is considered an indicator of both sympathetic and parasympathetic nervous system activity, and its ratio to HF, an indicator of parasympathetic nervous system activity, can quantitatively indicate whether the sympathetic or parasympathetic nervous system is dominant. Such analysis has been used clinically to understand the tension state of sympathetic and parasympathetic nervous systems [47,48]. HRV is an objective measure of stress in healthcare workers, and Joseph et al. [49] found self-reported stress to be associated with proportionately elevated physiological levels. Their results provided convincing evidence that physicians who routinely perform surgical tasks assess their own stress under severe time constraints. The dissemination of objective and ecologically valid measures of stress may offer important clues for understanding stressful situations and reducing the psychological load [50,51].
HRV measurements are derived from RR data and are influenced by the length of the time series (number of data points), body posture, and activity type. In the present study, these factors were derived using the five-minute RRI, which provides values for each activity. Fuentes-García et al. [52] reported lower average HRV values even during relatively short periods of mental stress.
Outliers or error beats due to artefacts are typically found in the time series of heartbeat intervals and have no physiological significance. Artefacts can significantly distort measurements in the time and frequency domains and increase the power over a wide band of frequencies [53]. In HRV data, if the time series is clean and sufficiently long to calculate the power of a given frequency band, the value is valid for evaluating the power of LF and HF. For example, at least 2.5 min of clean data are required to evaluate LF power [54]. For a stable interpretation of autonomic function, Chen et al. [55] established that time-domain HRV indices (e.g., SDNN, RMSSD, pNN50) require one minute of short-term recordings, while the specifications for frequency-domain HRV indices (LF and HF) require at least three minutes of recorded data to be accurately measured. The HRV data analysed in this study comprised one segment of five minutes, which is considered sufficiently long.
The data set collected in this study consisted of evaluation data collected at five-minute intervals and was filtered based on previous studies to exclude the effect of artefacts on HR [56,57]. Briefly, at the centre point of a moving window of length l, data points outside the interval were excluded, and the mean of the data points within the moving window was calculated, excluding the centre point; a is a positive number equal to or less than one. In this study, l = 41 and a = 0.2 were used.

Workload (Percentage HR Reserve) and PA
To determine the physical load during waste management work, the HR and PA of workers were used as indicators. The basic relationship equations are presented in Appendices E and F. Percentage heart rate reserve (%HRR) is a measure of the physical load or pressure intensity associated with muscle activity [58]. Norton et al. [59] stated that 40-60% HRR lasting 30-60 min is equivalent to moderate physical load. Hwang and Lee [16] and Hashiguchi et al. [10], who focused on construction workers, noted that an HRR of 30-40% in all-day work continues to pose a health risk for workers. Equation (1) shows how it can be estimated: where HR working is the average working heart rate, HR resting is the resting heart rate, and HR maximum is the maximum heart rate based on age [60]. In this study, HR resting was defined as the lowest stable five-minute heart rate in the work break. A small accelerometer (WHS-2) attached to the sensing wear placed on the chest was used to obtain the composite acceleration along three axes (vertical axis: X, horizontal axis: Y, and vertical axis: Z) [23]. Five-minute average values were used to calculate the amount of PA and to observe the overall intensity of workers' movements during working hours. The intensity of PA was calculated from the five-minute average values, which is a predictor of health status [61]. Detailed movement data could be obtained from the biometric data parameters, thus providing an understanding of the movement intensity of workers concerning work actions performed during the working day (e.g., lifting a load, carrying a load and walking, standing, and squatting) [62].

Data Collection
We collected data on the participants' HR, physical acceleration, and body surface temperature at various times during the working day. The WBGT of each group was measured in their respective work areas. A total of 1945 data points, measured approximately every five minutes, were collected from the 29 participants. The monitoring of workers was conducted throughout the working day, which included a one-hour break, and observations were made from 9:00 a.m. to 4:30 p.m. The data are presented in Appendix D. We collected self-reported information on age, years of work experience, height, and weight from the workers.
The research team did not monitor participants' behaviour but recorded their work activities using two cameras installed near each work area. Prior to the experiment, we told the participants that we would not measure their operating skills and instructed them not to deviate from their daily working routines, thereby avoiding the Hawthorne effect [63].

Descriptive Statistics and Intergroup Comparisons
The normality of data was tested using the Shapiro-Wilk and Kolmogorov-Smirnov tests (Appendix G). In both tests, the null hypothesis assumes that the data set is normally distributed at p = 0.05 [64]. However, there was a slight discrepancy between the tests; all the data except HR satisfied the condition of normal distribution.
The Mann-Whitney U test does not require a normally distributed data set [65], and its null hypothesis is 'no difference between the two groups at a significance level of 0.05.' At p < 0.05, the null hypothesis is rejected, indicating a statistically significant difference in the distribution of data for each group [39]. Statistically significant differences were found between workers for PA, BS TEMP, WBGT, NN50, pNN50, and RMSSD of HRV time-domain parameters and LF power, HF power, and LF/HF of HRV frequency domain parameters. For all the other parameters, no statistically significant differences were found between workers. The results of the analysis are presented in Table 1.
In addition, the time-domain parameters related to the autonomic nervous system, NN50, pNN50, and RMSSD, were low in waste management workers, thus indicating a decrease in parasympathetic activity or a state of relaxation. In the frequency-domain parameters, they may also have lower LF power associated with the parasympathetic nervous system and higher LF/HF, indicating stress. The ratio of LF power to HF power (LF/HF) is a classic indicator of sympathetic balance [66]. This value represents the overall balance between the sympathetic and parasympathetic nervous systems. A high value indicates sympathetic dominance, while a low value indicates parasympathetic dominance. Note: Group A = waste treatment workers; Group B = non-waste treatment workers; EXP = experience; BMI = body mass index; HR = heart rate; %HRR = percent heart rate; PA = physical activity; BS TEMP = body surface temperature; WBGT = wet bulb globe temperature; HRV = heart rate variability; RRI = heart beat interval; SDRR = standard deviation of heart beat interval; CVRR = coefficient of variation of heart beat interval; NN50 = index of difference between adjacent normal heart rate intervals greater than 50 ms; pNN50 = index of the percentage of the difference between adjacent normal heart rate intervals greater than 50 ms; RMSSD = root mean square successive difference; HRV = heart rate variability; LF = low frequency; HF = high frequency. Among the parameters, only HR showed a normal distribution, so HR was t-tested. Other parameters were tested for significance using the Mann-Whitney U test.

Relationship of Workers' Characteristics, Body Load, and HRV Index with Body Surface Temperature
The data collected in each work environment were merged and analysed for their effects on the body surface temperature of the workers. Multiple regression analysis was performed to evaluate significant relationships between each parameter. The results are presented in Table 2. In the multiple regression analysis, we checked for multicollinearity among the independent variables (worker characteristics, body load, and HRV index). Age, EXP, and BMI for worker characteristics, HR and PA for workload, SDNN, CVRR, and RMSSD for HRV time domain, WBGT for the work environment, and LF power, HF power, and LF/HF for HRV frequency domain all had variance inflation factors (VIFs) of less than 10. Age, HR, PA, and WBGT showed a significant relationship with BS TEMP. Note: BS TEMP = body surface temperature; EXP = experience; BMI = body mass index; HR = heart rate; PA = physical activity; WBGT = wet bulb globe temperature; SDRR = standard deviation of normal heart rate interval; CVRR = coefficient of variation of heart beat interval; RMSSD = root mean square successive difference; HRV = heart rate variability; LF = low frequency; HF = high frequency; B = slope (coefficient) of regression equation.
As shown in Table 3a,b, higher HR and PA were positively correlated with BS TEMP. Table 3c,d reveals that WBGT and AGE had a positive effect on BS TEMP, but the coefficient of determination was low, which suggests a small but statistically significant effect.

Indexes of Physical and Psychological Load
The data collected from each work environment were merged to evaluate the relationship among worker characteristics, PA, BS TEMP, HRV index, and physical and psychological loads as dependent variables. Table 4 presents the results. Prior to the multiple regression analysis of workload, multicollinearity was assessed with VIF, and no multicollinearity among the independent variables was confirmed. Among the independent variables of workload %HRR, BMI, PA, BS TEMP, Age, EXP, and LF power had a VIF of one to two, suggesting no possibility of multicollinearity. Multiple regression analysis of psychological load LF/HF indicated that among the independent variables of psychological load LH/HF, the indices RRI, PA, BS TEMP, Age, experience, and LF power had low VIF, thus suggesting no possibility of multicollinearity. RRI in workload %HRR and BMI in psychological load LH/HF were excluded from the independent variables in each multiple regression equation because VIF > 10. For each workload %HRR and psychological load LF/HF ratio, statistically significant independent variables were adopted in the multiple regression equations.

Relationship to Workload %HRR and Psychological Load LF/HF
First, we obtained a multiple regression equation that demonstrated the effects of the independent variables PA, BS TEMP, and age on the workload. The regression equation, as shown in Equation (2), has an adjusted R 2 of 0.449. The results of the analysis are shown in Table 5  Subsequently, we obtained a multiple regression equation showing the effects of the independent variables RRI, BS TEMP, LF power, and EXP on psychological load. The multiple regression equation, as shown in Equation (3), has an adjusted R 2 of 0.356. The results of the analysis are given in Table 6, wherein changes in BS TEMP and LF power had a positive impact on the psychological load of workers, whereas RRI and EXP had a negative impact. The standard regression bias of the independent variables suggested that the greatest effect was due to the LF power related to the autonomic nervous system. The biometric data of waste treatment workers were analysed to obtain a multiple regression equation for estimating their load at work, and the results established that BS TEMP was significantly affected by both workload (%HRR) and psychological load (LH/HF).

Principal Findings
This study investigated the effects of workers' individual characteristics and biometric indices on their physical and psychological loads during waste-handling operations. The results indicated significant differences in the workers' HRV time-domain parameters (NN50, pNN50, and RMSSD), and HRV frequency parameters (LF power, HF power, and LF/HF) when comparing waste handling and non-waste handling operations. Compared with workers who did not handle waste, those who did had a lower PA and BS TEMP; lower HRV parameters (NN50, pNN50, RMSSD) associated with higher psychological load, and higher LF/HF and LF power were associated with stress. There were no significant differences in HR, RRI, and %HRR. These results suggest that waste disposal workers work under a high psychological load, regardless of their HR and workload %HRR due to their work.
Workers' workloads and psychological loads were estimated using measurement parameters. Although studies have analysed the relationship between body temperature, WBGT, and HR [67,68], this study confirmed that body surface temperature could be used for this estimation. Workload %HRR was estimated from PA, body surface temperature, and workers' age (adjusted R 2 = 0.449), and psychological load LF/HF was estimated from the autonomy-influenced RRI, LF power, BS TEMP, and years of experience (adjusted R 2 = 0.356). PA and BS TEMP had a positive effect on workload %HRR [17], and the three independent variables, including workers' age, affected workload almost equally. RRI had a negative effect, and LF power and BS TEMP had a positive effect on the psychological load of LF/HF of workers [69,70]. There was a tendency to report a lower psychological load with increasing years of work experience. These findings support all three hypotheses. Although RMSSD, NN50, and pNN50 showed significant differences when comparing workers who handled waste to those who did not, they were not parameters that determined psychological load LF/HF. Orsila et al. [38] confirmed the relationship between RMSSD and mental stress during relatively long periods of time in the morning, afternoon, and evening among employees of an electronics company. However, the environment and conditions of work were different compared with this study, which focused on waste work, and there were differences in the results of each of these studies.

Physical and Psychological Load in Waste Management Workplaces
UEMAE et al. [71] observed the relationship of temperature and humidity near the skin to changes in HR, a physiological index, and stated that changes in skin temperature affect HR and that the factors are related to HR's autonomous thermoregulatory function [71]. The sympathetic activity of the autonomic nervous system can be used to estimate the intensity of physical and psychological loads, such as stress [72]. Therefore, it can be inferred that the body surface temperature affects %HRR and HRV, which are load indicators related to the work environment. This inference is based on the relationships shown in Equations (2) and (3). We also found that waste workers are exposed to various levels of stress due to the handling of waste and the temperature of their work environment.
Our results indicate that uncertainty about workers' psychological load and its acceptable limits can be resolved by analysing the environment in which waste is handled and the temperature conditions of the workers.
In the estimation of %HRR, body surface temperature was affected as an explanatory variable, but there was no effect of environmental temperature on WBGT. This relationship indicates a low correlation between body surface temperature and WBGT, as evident from Table 3c. Studies have reported the effects of WBGT on biological information [18,73]. However, it can be inferred that WBGT near each work site may not be related to workers' body temperature and that analyses using body surface temperature can show changes in worker HR more accurately. Furthermore, the LF/HF estimation revealed that workers handling waste had higher sympathetic nervous activity (i.e., higher LH/HF and LF power) than parasympathetic nervous activity (i.e., lower RRI, SDRR, RMSSD, and HF power) related to the catatonic state. In addition, workers' HRV indices were low, which aligned with previous findings [74,75], suggesting that increased work-related stress leads to decreased parasympathetic activity and increased sympathetic activity. Thus, the LF/HF ratio of workers provides insight into their work stress levels.
The participants of this study were exposed to low physical load and high mental stress. This indicates that waste workers may experience higher psychological stress due to the characteristics of their jobs, such as taking extra care in regards to the safety of their environment and working with hazardous waste, than those engaged in hard physical labour. Eggenberger et al. [30] demonstrated that the frequency of exposure of waste handlers to hazardous materials significantly increased perceptual stress, and workers with a higher frequency of exposure reported 2.14 times more stress than those with a lower exposure. As waste management companies have high psychological job demands, it is necessary to assess workers' job-related psychological factors. It would be useful for waste management companies to use this research tool to regularly and quantitatively assess the physical and psychological load of workers. Instead of using questionnaires and interviews to assess workers' health and well-being, this research tool will enable companies to more effectively assess workers' health and provide an appropriate working environment.

Approaches to Worker Stress Reduction and Worker Management
In the estimation of workers' psychological load, experience was inversely correlated with LF/HF (standard β of EXP = −0.018). Experience is negatively correlated with LF/HF because, as workers gain more experience, they have more knowledge about dealing with the waste. Thus, the psychological load may be reduced by improving inexperienced workers' knowledge of health literacy, waste to be handled, protective gear, and tools and education and training. Studies using questionnaire surveys of waste workers have established that worker demographics (e.g., education level, marital status, and the number of children) and working conditions (e.g., shift work, work hours, and income) are significantly related to psychological load levels [76][77][78]. Waste workers have a high turnover rate, which makes it difficult to collect personal information in some cases. Our results confirm that work stress can be estimated independent of worker demographics and working conditions, and may be used for worker management independent of country or industry.

Theoretical and Practical Contributions
This study makes the following theoretical contributions. The psychological load LH/HF by HRV and %HRR as workload were used to quantify the workload of waste management workers. In addition, equations for estimating these loads using physical and psychological loads from HR and biometric data, including PA and body surface temperature, were developed.
Regarding practical contributions, the study found that the physical and psychological loads of waste workers, which have not been clarified previously, differ depending on the type of work. This evidence and monitoring of workers will assist the waste management industry to reduce health risks and enhance the working conditions of workers, and it can suggest new management concepts to companies by considering these factors from the perspective of health psychology.

Limitations
This study has several limitations. First, physical characteristics and age were not equally balanced in the sample. Although physical characteristics affecting workload and stress vary with age and gender [79,80], the results may not be generalisable to all waste workers as the study was conducted on healthy men aged 20-63 years. The waste management industry in Japan is dominated by men, and future studies should include women in their sample. Second, this study was a cross-sectional analysis, with data collected during the working days for seven days. Additionally, the mental and physical status of workers at the beginning of the workday was not measured. Since nutritional status, family environment, past stress conditions, lack of sleep, and lack of physical activity (sports) affect personnel activities, observing workers' conditions over a longer period of time could yield more useful results. Investigation of background factors and initial conditions should be the subject of future research. Finally, as frequency-based metrics have been reported to represent the balance between sympathetic and parasympathetic activity more accurately [81], it is vital to improve the quality of HR interval recordings in wearable devices. Certain data collected in this study had missing HR data intervals, which affected the selection of HRV metrics and necessitated the removal of a few participants from the statistical analysis. Future studies aimed at further improving sensing wear and wearable technology and enhancing recording quality (e.g., further minimising motion artefacts) are essential to investigate large samples over time. Such improvements will increase the usefulness of these devices.

Conclusions
In this study, physical and psychological loads in the workplace were investigated among workers at a waste management facility in Japan. The workers performed their daily tasks in an environment exposing them to high stress depending on environmental temperature and task. Those who directly handled waste had lower PA, lower body surface temperature, and lower HRV parameter values associated with the autonomic nervous system than maintenance workers at the facility. Workload and stress among waste management workers, which has not been investigated in previous studies, suggested a significant relationship between HR variability, body surface temperature, PA, age, and years of experience. The workers' body surface temperature was significantly related to HR, PA, environmental temperature, and age. The psychological load LF/HF was affected by the HRV parameters, RRI and LF power, body surface temperature, and years of experience.
Regular monitoring of biometric and physical information such as HR, body surface temperature, and activity can help understand the working environment of workers with high physical and psychological loads and promote the maintenance of sustainable relationships among them in harsh manufacturing and construction environments [10]. A quantitative understanding of working conditions in harsh work environments is important for improving work quality and reducing health hazards. Waste management companies can reduce the negative impacts on the environment and public health by promoting sustainable working relationships.

Acknowledgments:
We would like to express our gratitude to the Kyoto Environmental Conservation Corporation who provided us with a working site for the experiment, and the waste workers who participated in the survey.

Conflicts of Interest:
The authors declare no conflict of interest.

Appendix B
Picture of measurement equipment for physiology in this study: (a) Heart rate and acceleration sampling device (WHS-2), (b) Sensing-ware (COCOMI ® ), and (c) Device for collecting data (CC2650). Electrodes are placed inside the COCOMI ® to acquire heart rate information and body temperature via the WHS-2.

Appendix B
Picture of measurement equipment for physiology in this study: (a) Heart rate and acceleration sampling device (WHS-2), (b) Sensing-ware (COCOMI ® ), and (c) Device for collecting data (CC2650). Electrodes are placed inside the COCOMI ® to acquire heart rate information and body temperature via the WHS-2. A X n 2 + A Y n 2 + A Z n 2 mG Appendix D