Empirical prediction model of psychophysiological responses of workers with respect to noise exposure based on random forest

Abstract

Evidence of the non-auditory effects of noise exposure on workers is growing. This study aimed to empirically present prediction models of psychophysiological responses of workers with respect to noise exposure. In this study, 169 male workers from typical industrial workrooms were asked to judge the mental workload, noise annoyance, and noise sensitivity during exposure to noise. Two main physiological responses that include heart rate and blood pressure were also measured. Noise exposure characteristics were measured using the calibrated instruments. Empirical prediction models were developed based on Random forest compared with the regression method. 46% of the workers were exposed to noise up to the exposure limit (85 dB for 8 h) and 54% of the workers were exposed to noise upper than it. It is observed that the considerable body response changes of workers exposed to noise from the medium to high levels (p < 0.05). Random forest could provide more accurate predictions than multiple regressions (R square = 0.73 to 0.80). Four variables as noise dose, noise sensitivity, age, and noise frequency are found to be the important factors influencing the psychological responses, respectively. Moreover, the main variables as noise sensitivity and noise dose and age are found to be the important factors influencing the physiological responses, respectively. Changes in the psychophysiological responses above the medium noise levels confirmed that the action level (82 dB for 8 h) can be a suitable criterion to prevent possible auditory and non-auditory complications. These dose-response models can be helpful in setting definitive exposure limits for noise-induced non-auditory effects at workplaces.

Keywords

Random forest noise exposure prediction model psychophysiological responses workers

Introduction

Noise is considered to be the most persistent physical contaminant in industrial workplaces for workers to expose. Noise exposure at levels even below hearing damage may cause body psychological and physiological changes. Based on observed auditory effects, most countries set the occupational exposure limit (OEL) of 85 dBA for an 8-hour time-weighted average exposure. It seems that the noise-induced non-auditory effects may be observed even at levels below the occupational exposure limit. Hence, based on the possible auditory and non-auditory responses, it is observed that an 8 h time-weighted action level criterion of 82 dBA, which corresponds to a dose of 50% of the OEL, is recommended for noisy workrooms. The wide range of non-auditory effects has led researchers to believe that noise can act as a non-specific stressor.¹ Cardiovascular effects, biological rhythm disruption, and simple physiological reactions may result from exposure to medium to high levels of occupational noise.² Moreover, the most common occupational noise-induced perceptual and behavioral symptoms that workers subjectively feel are mental workload, annoyance, and fatigue.³ Noise can affect basic mental activities such as perception, attention, concentration, and communication, as well as annoyance.⁴ Noise annoyance is influenced by the humans' sensitivity characteristics,⁵ so that when highly sensitive individuals are exposed to noise, they get more annoyed.^6,7 Noise annoyance is the most prevalent community response in workers exposed to noise.⁸ Mental workload refers to a person’s mental ability. The extra costs levied on the operator due to noise to reach a certain degree of efficiency are known as noise-induced mental workload.⁹ The noise level is described as a key factor influencing these responses. The adverse effects of noise are often determined by the noise intensity, frequency, and exposure time.¹⁰ Some personal variables such as age, work experience, general health status, work pattern, individual susceptibility, tobacco use, and personal characteristics can also affect these outcomes.⁴ Golmohammadi et al. claimed that noise sensitivity is one of the most important predictors of the psychological effects of noise.¹¹ It has been reported that acute noise exposure causes some physiological responses.¹² These changes can result in deregulation and long-term physiological changes, raising the risk of cardiovascular disease.⁴ However, Fyhri et al. stated that there is no correlation between noise exposure and cardiovascular responses.⁴ Jarup et al. found a significant relationship between aircraft noise and hypertension in those who were exposed to nighttime noise.¹³ Veljovic et al. stated that it is possible to predict body blood pressure for a period when exposed to a certain level of noise.¹⁴

Despite the focus on exposure limits for noise to prevent hearing loss, there is restricted information about noise exposure limits to prevent non-auditory effects. Epidemiological studies have also not presented an accurate pattern between occupational noise exposure and non-auditory responses. The risk assessment of the non-auditory responses of workers exposed to noise has become an important topic worldwide. These effects are still questionable for realistic situations with long-term noise exposure. Moreover, less attention has been paid to workers’ non-auditory responses due to noise doses from medium to high levels at industrial workplaces.

It seems that for accurate analysis of dose-response about the non-auditory effects of noise, more advanced statistical methods are to be needed. Currently, machine learning is recognized as a new method for modeling various sciences, accurately with the ability to determine the complex relations among different variables. Random Forest is a powerful technic employed widely across a multitude of fields. Random forest model is a flexible, easy-to-use machine learning algorithm that generates, even without hyper-parameter tuning, a great output most of the time. The variable importance can help with a better understanding of the solved problem. A great quality of the random forest algorithm is that it is very easy to measure the relative importance of each feature on the prediction.¹⁵

This study aimed to develop an empirical model for analyzing psychological and physiological responses in workers concerning to noise exposure using a machine learning algorithm. Moreover, multiple linear regression was also employed and their results were compared with those of machine learning. It is expected that this model will be able to determine the relative importance of the individual risk factors on physiological and psychological responses.

Methods

Experiment design

This cross-sectional study was conducted in Hamadan city located in the west of Iran in three typical industries in 2020 which workers have noise exposure from medium to high levels. Figure 1 showed the flowchart of different steps for the development of the prediction models. The main steps included were selection of the participants, measurement of baseline body responses before starting working time, measurement of noise exposure during work, and re-measurement of body responses at the end of working time.

Figure 1.

Flowchart of different steps for development of the prediction models.

Source of data

In the study, 169 male workers with good general health and normal hearing were randomly participated. Worker demographic information including age, work history, marital status, body mass index (BMI), and smoking was collected based on the questionnaire. Inclusion criteria were at least 1 year of employment in the current job and no special physiological and psychological diseases. The night before each session, workers were asked to get plenty of sleep. Before starting the work, written consent was received from the workers and they were informed about the aim and method of the study. This study was approved by the Ethics Committee of Hamadan University of Medical Sciences (Ethics Code: IR. UMSHA.REC.1398.906). The workers were selected from workstations where noise exposure was from medium to high levels. It should be noted that exposure to other main environmental factors such as lighting, air temperature, and chemical agents was relatively the same in the studied workstations. During the data collection, workers were placed away from their workstation and were moved to a quiet location.

Noise exposure

According to the International Standard Organization (ISO 9612) standard,¹⁶ equivalent noise levels for eight working hours were measured using a noise dosimeter (TES-1354, TES Electrical Electronic Corp) at each workstation during normal shiftwork. Noise frequency analyses were also conducted using a sound level meter (SVAN 971, SVANTEK Electronics manufacturer) which was calibrated using a calibrator. Noise dose is defined as the A-weighted equivalent noise level (Leq), to which a worker may be subjected for a normal working day of 8 h. The advantage of expressing the noise dose in this manner is that 100% will always represent the criterion dose whatever the measurement duration and however it is accumulated. As mentioned according to the legislation criterion, equivalent noise exposure of 85 dB for 8 h is equal to 100% noise dose or a man exposed to 82 dB for 8 h has a 50% noise dose. Noise exposures based on the dominant frequency spectrum were classified into three categories including low-frequency noise (lower than 250 Hz), medium frequency noise (from 500 Hz to 1000 Hz), and high-frequency noise (upper than 2000 Hz). To prevent the sources of uncertainty due to variations in daily work, long-duration measurements using personal noise dosimeters were conducted. Some repeat noise measurements were also performed randomly to ensure the data’s reliability and accuracy.

Psychological responses

Noise annoyance (NA), mental workload (MW), and noise sensitivity (NS) were selected as psychological response criteria. At the end of each shiftwork, the psychological responses were measured. The noise-induced annoyances of the subjects were assessed using a numerical rating scale proposed by ISO 15,666.¹⁷ There are several methods for assessing the mental workload that included the NASA Task Load Index (TLX), the Subjective Workload Assessment Technique (SWAT), and the Workload Profile (WP) methods.¹⁸ In this study, the SWAT method which has a short response time was used to assess time load, mental effort load, and stress load. The scoring phase is usually completed during the task and is a quantitative ranking strategy that uses three stages for each of the three dimensions of time load, mental effort load, and psychological stress load: (1) low, (2) medium, and (3) high.¹⁸ The workers were asked to complete noise sensitivity (NS) based on the Weinstein noise sensitivity questionnaire.¹⁹ This questionnaire has 21 questions in 4 subscales, noise sensitivity, and the impaired concentration of questions, attitude to noise in the living location, and attitude to noise control and it is completed in a range of Likert scale so that the scores below 75 are considered as low sensitivity.

Physiological responses

Two physiological responses including heart rate (HR) and systolic blood pressure (SBP) were measured before starting work as a baseline and at end of working time. For each measurement, 10 min before the measurement, workers were asked to rest on a chair based on the World Health Organization recommendations for taking blood pressure.²⁰ In the seated position, workers’ blood pressure and heart rate in the right arm were measured three times at one-minute intervals. The mean of the second and third measurements was considered as the subject’s blood pressure and heart rate.²¹ Then, systolic blood pressure and heart rate were recorded by a trained technician using an automatic arm blood pressure monitors. To reduce uncertainty, instead of using the observer’s hearing, the value of blood pressure for each worker was calculated using the numbers shown on the tool’s monitor screen, meaning that our observations were not impacted by random errors in the observer’s hearing. The experiment conditions were the same for all participants. The physiological response measurements were repeated three times to ensure reproducibility.

Model development

Description of model variables

In the prediction models for body psychological responses, based on the random forest and the multiple linear regression, eight input variables including age, married status, work experience, BMI, smoking status, noise frequency, noise sensitivity, and noise dose were considered. In the prediction models for body physiological responses based on the random forest and the multiple linear regression, 10 main input variables including age, married status, work experience, BMI, smoking, baseline BP, baseline HR before, noise frequency, noise sensitivity, and noise dose were considered.

Random forest model

The random forest as a supervised learning algorithm can be used for both classification and regression problems. The forest is an ensemble of decision trees, always trained with the “bagging” method. The basic idea of the bagging method is that a combination of learning models increases the accuracy of the overall result.²² Random Forest includes training each decision tree on a different data sample where sampling is done with replacement. In the setting of the regression problem, the method operates by constructing a multitude of decision trees at the training phase and outputting the mean/average prediction of the individual trees.¹⁵

Random forests can be used to rank the relative importance of variables in a regression or classification problem. Calculation of the variable importance is performed by looking at the change in prediction error occurring when out-of-bag data for that variable is randomly permuted while all other variables are left unchanged. The calculations are performed tree by tree while the RF is drawn. Compared to variables that are not important, permuting values of an important variable in the analysis problem at random leads to greater changes in prediction performance.²³

The random forest models were developed with input and output features as multiple linear regression. We used default parameters for Random forest: the number of trees (ntree) equal to 1000 and the number of variables analyzed at each node to find the best split where the total number of variables in the problem is.²⁴

Multiple linear regression (MLR)

To investigate the effects of exposure to noise on body psychophysiological responses, a multiple linear regression was developed. The purpose of MLR is to establish the equation that expresses psychophysiological response changes as a function of noise exposure characteristics.²⁵ Multiple regression model (MLR) is presented as shown in equation (1)

Y = b_{0} + b_{1} \times x_{1} + b_{2} \times x_{2} + \dots \dots b_{p} \times x_{p} + ε (1)

(1)

where Y represents the predicted value of the dependent variable, x represents the predictor variables, b represents the slope, and ε is the random error.

Model evaluation criteria

The predictive performance of random forest and multiple linear regression models was evaluated based on the RMSE (root mean square error) and R square. Statistical analyses were performed using R Statistics Packages.

Results

Descriptive of the model variables

The results of noise dose are presented in Table 1. The results indicated that the workers were in exposure to the equivalent noise levels of 86.3 ± 2.1 dB (min = 83.4 dB, max = 91.5 dB). In the current study, 46% of the workers were exposed to noise from the medium noise levels to the noise exposure limit (85 dB for 8 h, noise dose of 100%). Moreover, 54% of the workers were exposed to noise upper than the mentioned exposure limit.

Table 1.

Descriptive statistics of the input and output variables for developing models.

Variable	Unit	Minimum	Maximum	Mean	Std. Deviation
Age	Year	23	54	36.87	7.32
Work experience	Year	2	25	8.68	5.94
BMI	Kg/m²	17.3	32.11	24.69	2.92
BP baseline	mmHg	90	158	117.58	11.34
HR baseline	bpm	50	95	74.17	7.98
Noise sensitivity	Score	26	84	44.36	11.12
Noise dose	%	68	412	154.82	84.92
Mental workload	Score	29	78	45.74	11.90
Noise annoyance	Score	3	10	5.72	1.91
Blood pressure	mmHg	95	172	124.56	12.70
Heart rate	bpm	58	100	79.37	7.83

The results showed that 17.2% (n = 29) of the workers were single and 82.8% (n = 140) were married. Also, 146 workers were non-smokers, and 23 of them smoked. 37.3% (n = 63) of workers exposed to the low-frequency noise, 32.5% (n = 55) of workers exposed to the medium frequency noise, and 30.2% (n = 51) of workers exposed to the high-frequency noise. Descriptive statistics of the main quantitative input and output variables for developing empirical models are reported in Table 1.

Descriptive statistics of body response variables based on the current OEL, equivalent noise exposure of 85 dB for 8 h equal to 100% noise dose, are presented in Table 2. Significant differences were observed in body psychological responses based on the occupational exposure limit. However, a significant difference was not observed about blood pressure response. The results generally showed that body response variables for the workers exposed to noise higher than the OEL were greater than the workers exposed to noise below the OEL (p < .05).

Table 2.

Descriptive statistics of the response variables based on the occupational exposure limit.

Output variable	Noise dose (%)		p-Value
Output variable	⩽100	>100
Heart rate (bpm)^a	4.20 ± 4.60	6.00 ± 5.10	<0.05
Blood pressure (mm Hg)^a	6.80 ± 5.10	7.07 ± 7.70	>0.05
Mental workload (score)	38.40 ± 5.10	52.2 ± 12.3	<0.05
Noise annoyance (score)	4.34 ± 0.85	6.90 ± 1.70	<0.05

^aThe values of heart rate and blood pressure are based on the mean difference before and after daily noise exposure.

Scatter plots of the body response variables of workers with respect to the noise dose are shown in Figure 2. The results showed that noise doses have some nonlinear trends with body psychological and physiological response indicators. However, the results indicated that an increase in noise dose from medium to high levels is associated with an increase in psychophysiological responses among workers (p < .05). Body psychological responses showed a greater relationship with respect to noise dose.

Figure 2.

Scatter plots of the psychophysiological responses of workers compared to the noise dose.

Prediction model results

The detailed coefficients of the developed multiple linear regression model for psychological and physiological responses are presented in Table 3. The purpose of the MLR was to establish an equation that expresses the psychophysiological response changes as a function of the noise exposure. The results showed that the relationship between the noise dose and all psychophysiological responses of workers in the presence of other factors was statistically significant (p < .05). As shown in Table 3, noise sensitivity is another main risk factor that was significantly associated with changes in all psychophysiological responses (p < .05).

Table 3.

Multiple linear regression model for psychophysiological responses with respect to noise exposure.

Variable	Mental workload (score)		Noise annoyance (score)		Heart rate (bpm)		Blood pressure (mmHg)
Variable	Beta	p-Value	Beta	p-Value	Beta	p-Value	Beta	p-Value
(Constant)	—	0.005	—	0.123	—	0.012	—	0.138
Age	−0.015	0.119	−0.051	0.435	0.014	0.829	−0.001	0.985
Work experience	−0.102	0.779	−0.013	0.800	−0.065	0.206	0.075	0.107
Noise dose	0.390	0.000	0.619	0.000	0.175	0.019	−0.028	0.048
Noise sensitivity	0.624	0.000	0.265	0.002	0.114	0.015	0.240	0.001
Smoking status	0.032	0.407	0.058	0.210	0.077	0.095	0.058	0.159
Married status	−0.062	0.146	−0.043	0.394	0.025	0.615	0.038	0.406
Dominant frequency	−0.071	0.084	0.032	0.514	−0.153	0.102	−0.067	0.121
BMI	0.023	0.555	0.053	0.250	0.051	0.282	−0.037	0.394
Baseline heart rate	—	—	—	—	0.777	0.000	−0.037	0.362
Baseline blood pressure	—	—	—	—	0.039	0.395	0.834	0.000
Model accuracy	R square = 0.79RMSE = 5.50		R square = 0.68		R square = 0.70		R square = 0.76
Model accuracy	R square = 0.79RMSE = 5.50		RMSE = 1.0		RMSE = 5.50		RSMSE = 6.30

RMSE, root mean square error: BMI, body mass index.

The performance of the developed random forest models for the prediction of the psychological and physiological responses is presented in Table 4. The performances of these empirical models based on the prediction criteria are at an acceptable level. The results showed that compared with multiple regression techniques, random forest models provide more accurate prediction.

Table 4.

Performance of the random forest models for the prediction of the psychological and physiological responses.

Output variable	R square	RMSE
Blood pressure (mm Hg)	0.80	5.0
Heart rate (bpm)	0.77	4.50
Noise annoyance (score)	0.73	0.98
Mental workload (score)	0.80	4.2

RMSE, root mean square error.

Relative importance of variables

As mentioned, the random forest algorithm has great potential to measure the relative importance of each feature on the prediction. Figure 3 showed the relative importance of input variables on psychological responses, noise annoyance, and mental workload based on random forest models. Among the input variables, three features as noise dose, noise sensitivity, and age are found to be the most important factors influencing the psychological responses, respectively. Noise frequency type was also recognized as another important variable. The effect size of other input variables can be observed in Figure 3. Some individual characteristics including married status, smoking status, BMI, and work experience are found to be less effective parameters.

Figure 3.

Relative importance of variables on noise annoyance (left) and mental workload (right) based on the random forest.

Figure 4 showed the relative importance of input variables on physiological responses, heart rate, and blood pressure based on random forest models. Among the input variables, some features as baseline values, noise sensitivity, and noise dose and also age are found to be the most important factors influencing the physiological responses, respectively. Based on the observed effect size, some factors including married status, smoking status, BMI, and noise frequency are found to be less effective factors on physiological responses.

Figure 4.

Relative importance of variables on blood pressure (right) and heart rate (left) based on the random forest.

Discussion

The body psychophysiological responses trend with respect to noise exposure, to date, is considered as the main challenge associated with non-auditory effects of noise pollution particularly in the industrial workplace. The current study empirically tries to explore some aspects of non-auditory effects of noise and the main risk factors associated with it employing the newly available statistical methods.

In this study, the results confirmed that the studied workers were in exposure to from medium level of 83.4 to a high level of 91.5 dB. Half of the subjects were exposed to noise upper than the current exposure limit so that the body responses for these workers were significantly greater than those exposed to noise below the OEL. It is observed that an increase in noise dose from medium to high levels is associated with a considerable increase in psychophysiological responses among workers. As mentioned, the action level of 82 dB (A) for 8 h (noise dose of 50%) is proposed as noise exposure criterion for the protection of the workers from the possible auditory and non-auditory effects of noise during working time. Observing changes in the body psychophysiological responses above the medium noise levels confirmed that the recommended action level (82 dB for 8 h) can be a suitable and applicable threshold for starting health care actions to prevent auditory and non-auditory complications among workers in long-term exposure.

It is observed that the developed MLR to express the psychophysiological responses changes as a function of the noise exposure showed that the relationship between the noise dose and all psychophysiological responses of workers was statistically significant. Noise sensitivity was another main risk factor that was significantly associated with changes in all psychophysiological responses.

The relative importance of input variables based on the random forest model confirmed that four variables as noise dose, noise sensitivity, age, and noise frequency are found to be the important factors influencing the psychological responses, respectively. Some variables as noise sensitivity and noise dose and also age are found to be the important factors influencing the physiological responses, respectively. However, noise frequency is found to be a less effective parameter on physiological responses. As expected, noise frequency was more influential on psychological responses than physiological responses. It should be noted that most of the studied workers are exposed to low-frequency noise which is recognized as responsible for many mental effects of noise exposure.

In line with the present study, Steinbach et al. also showed that the individuals’ irritation levels were significantly correlated with the noise dose and smoking, noise sensitivity, and dominant noise frequency.²⁶ Moreover, people who were neurotic and sensitive to the noise experienced more noise-induced annoyance. Many studies have shown that noise dose and noise sensitivity are closely linked to irritation.^27,28 Some studies showed exposure to noise levels above 55 dB could have intermittent health consequences with an increase in the percentage of annoyance in a large population.²⁹ Calderwood et al. showed that some individual characteristics affect individual reasons of state of mental fatigue.³⁰ Some studies showed noise sensitivity has an important impact on physiological responses.³¹ A cross-sectional study of 188 male workers in Taiwan found that exposure to 80 decibels for 2 to 4 years increased the risk of high blood pressure.³² Bjor et al. showed that exposure to noise will influence heart rate outcomes.³³ The current results also verified that among individual characteristics, workers’ age and noise sensitivity have the greatest impact on physiological and psychological responses. Therefore, the noise sensitivity variable can be used as one of the criteria for selecting healthy workers for noisy jobs and tasks.

Developing an empirical model is a common approach used to explore relations among some possible variables subjected to desired outputs, especially when it is difficult or impossible to develop a comparable mathematical model. The results showed that noise doses have some nonlinear trends with body psychological and physiological response indicators. In this regard, the results confirmed that compared with multiple regression techniques, random forest models have a more accurate prediction for body psychological and physiological responses.

It can be seen the high capabilities of machine learning algorithm to improve the prediction performance compared with current empirical techniques. Using the developed models, occupational health professionals can accurately analyze body psychophysiological responses of workers exposed to noise based on some simple environmental and body characteristics. However, the application domain of these empirical models was restricted to the range of input variables (min–max).³⁴ The main sources of predictive error in these empirical models include the type and the number of selected input variables. In real situations, there are many factors that could influence the body’s psychophysiological responses. However, all of them cannot be incorporated into the practical model. It should be noted that the level of detail of input variables should be in accordance with the desired value of the accuracy of predictions.³⁵ Despite the existing research studies related to exposure–response relationships of noise and non-auditory effects, these topics are needed for future empirical studies with a very large sample size based on the new advanced statistical methods. These new statistical techniques can be used to solve many problems in the topic of noise exposure risk assessment once thought impossible just a few years ago.

Conclusion

This research aimed to explore new evidence and provide useful empirical data on the psychophysiological responses of workers with respect to noise exposure, particularly in the industrial workplace. The finding confirmed that an increase in noise dose from medium to high levels is associated with an increase in psychophysiological responses. Body psychological responses of workers showed a greater relationship with respect to noise dose. Observing changes in the body psychophysiological responses above the medium noise levels confirmed that the recommended action level (82 dB for 8 h) can be a suitable and applicable threshold for starting health care actions to prevent auditory and non-auditory complications among workers in long-term exposure. Random forest models could provide more accurate predictions than multiple regressions. For noisy workrooms, it is recommended that sensitivity to noise and age restrictions should be considered in pre-employment examinations. These model outputs can be employed in setting exposure limits for noise-induced non-auditory effects at workplaces.

Footnotes

Acknowledgements

The authors would like to express their gratitude to all the workers who have participated in this research.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was financially supported by the Research Deputy of Hamadan University of Medical Sciences (Grant number: 9811299112).

ORCID iD

Mohsen Aliabadi

References

Christa

Elizabeth

. Occupational noise exposure: a review of its effects, epidemiology, and impact with recommendations for reducing its burden. J Acoust Soc Am 2019; 146(5): 3879–3905.

Brännström

Johansson

Vigertsson

, et al. How children perceive the acoustic environment of their school. Noise Health 2017; 19(87): 84–94.

Golmohammadi

Darvishi

Faradmal

, et al Attention and short-term memory during occupational noise exposure considering task difficulty. Appl Acoust 2020; 158: 107065.

Fyhri

Aasvang

. Noise, sleep and poor health: modeling the relationship between road traffic noise and cardiovascular problems. Sci Total Environ 2010; 408(21): 4935–4942.

Michaud

Keith

Feder

, et al. Personal and situational variables associated with wind turbine noise annoyance. The J Acoust Soc America 2016; 139(3): 1455–1466.

Abbasi

Yazdanirad

Habibi

, et al Relationship among noise exposure, sensitivity, and noise annoyance with job satisfaction and job stress in a textile industry. Noise & Vibration Worldwide 2019; 50(6): 195–201.

Gille

Marquis

Weber

. Aircraft noise annoyance modeling: consideration of noise sensitivity and of different annoying acoustical characteristics. Appl Acoust 2017; 115: 139–149.

Berglund

Lindvall

Schwela

. Guidelines for community noise. World Health Organization. The Stockholm University and Karolinska Institute, Sweden, 1999.

Darvishi

Maleki

Giahi

, et al. Subjective mental workload and its correlation with musculoskeletal disorders in bank staff. J Manipulative Physiol Ther 2016; 39(6): 420–426.

10.

Basner

Babisch

Davis

, et al. Auditory and non-auditory effects of noise on health. The Lancet 2014; 383(9925): 1325–1332.

11.

Golmohammadi

Darvishi

Motlagh

, et al Role of individual and personality traits in occupational noise-induced psychological effects. Appl Acoust 2021; 173: 107699.

12.

Haralabidis

Dimakopoulou

Vigna-Taglianti

, et al. Acute effects of night-time noise exposure on blood pressure in populations living near airports. Eur Heart J 2008; 29(5): 658–664.

13.

Jarup

Babisch

Houthuijs

, et al. Hypertension and exposure to noise near Airports: the HYENA study. Environ Health Perspect 2008; 116(3): 329–333.

14.

Veljovic

Burak

Begic

, et al .Noise effect on blood pressure and heart rate - Regression analysis in service of prediction. Acta Inform Med 2019; 27(3): 162.

15.

Biau

. Analysis of a random forests model. J Mach Learn Technol 2012; 13(1): 1063–1095.

16.

ISO 9612. Acoustics- determination of occupational noise exposure - Engineering method. International Organization for Standardization, Geneva, Switzerland, 2009.

17.

ISO 15666. Acoustics- assessment of noise annoyance by means of social and socio-acoustic surveys. International Organization for Standardization, Geneva, Switzerland, 2003.

18.

Rubio

Diaz

Martin

, et al. Evaluation of subjective mental workload: a comparison of SWAT, NASA-TLX, and workload profile methods. Appl Psychol 2004; 53(1): 61–86.

19.

Ekehammar

Dornic

. Weinstein’s noise sensitivity scale: reliability and construct validity. Percept Mot Skills 1990; 70(1): 129–130.

20.

Hackam

Quinn

Ravani

, et al. The 2013 Canadian hypertension education program recommendations for blood pressure measurement, diagnosis, assessment of risk, prevention, and treatment of hypertension. Can J Cardiol 2013; 29(5): 528–542.

21.

Eghbali

Khosravi

Feizi

, et al Prevalence, awareness, treatment, control, and risk factors of hypertension among adults: a cross-sectional study in Iran. Epidemiol Health 2018; 40: e2018020.

22.

Khalilia

Chakraborty

Popescu

. Predicting disease risks from highly imbalanced data using random forest. BMC Med Inform Decis Mak 2011; 11:1–13. DOI: 10.1186/1472-6947-11-51.

23.

Shaikhina

Lowe

Daga

, et al. Decision tree and random forest models for outcome prediction in antibody incompatible kidney transplantation. Biomed Signal Process Control 2019; 52: 456–462.

24.

Alam

Rahman

. A Random Forest based predictor for medical data classification using feature ranking. Inform Med Unlocked 2019; 15: 100180.

25.

Yılmaz

Yuksek

. An example of artificial neural network (ANN) application for indirect estimation of rock parameters. Rock Mech Rock Eng 2008; 41(5): 781–795.

26.

Steinbach

Altinsoy

. Prediction of annoyance evaluations of electric vehicle noise by using artificial neural networks. Appl Acoust 2019, 145: 149–158.

27.

Ali

. Industrial noise levels and annoyance in Egypt. Appl Acoust 2011; 72(4): 221–225.

28.

Brocolini

Parizet

Chevret

. Effect of masking noise on cognitive performance and annoyance in open plan offices. Appl Acoust 2016; 114: 44–55.

29.

Hume

. Sleep disturbance due to noise:Current issues and future research. Noise Health 2010; 12(47): 70.

30.

Calderwood

Ackerman

PL.

The relative impact of trait and temporal determinants of subjective fatigue. Pers Individ Differ 2011; 50(4): 441–445.

31.

Park

Lee

Jeong

. Effects of noise sensitivity on psychophysiological responses to building noise. Build Environ 2018; 136: 302–311.

32.

Chang

Liu

Young

, et al. Noise frequency components and the prevalence of hypertension in workers. Sci Total Environ 2012; 416: 89–96.

33.

Björ

Burström

Karlsson

, et al. Acute effects on heart rate variability when exposed to hand transmitted vibration and noise. Int Arch Occup Environ Health 2007; 81(2): 193–199.

34.

Aliabadi

Golmohammadi

Ohadi

, et al Development of an empirical acoustic model for predicting reverberation time in typical industrial workrooms using artificial neural networks. Acta Acustica United with Acustica 2014; 100(6): 1090–1097.

35.

Shafiee Motlagh

Golmohammadi

Aliabadi

, et al. Empirical study of room acoustic conditions and neurophysiologic strain in staff working in special open-plan bank offices. Acoust Aust 2018; 46: 329–338.