• Users Online: 192
  • Print this page
  • Email this page


 
 
Table of Contents
ORIGINAL ARTICLE
Year : 2020  |  Volume : 9  |  Issue : 1  |  Page : 23-27

A new approach to detect the physical fatigue utilizing heart rate signals


1 School of Architecture, Islamic Azad University Taft, Iran
2 School of Architecture and Built Environment, Deakin University, Geelong, Australia
3 System Administrator, Dibrugarh University, Assam, India
4 Institute for Intelligent Systems Research and Innovation, Deakin University, Geelong, Australia
5 Rajaie Cardiovascular Medical and Research Center, Iran University of Medical Sciences, Tehran, Iran

Date of Submission17-Mar-2020
Date of Decision20-Mar-2020
Date of Acceptance30-Mar-2020
Date of Web Publication24-Apr-2020

Correspondence Address:
Dr. Zahra Alizadeh Sani
Rajaie Cardiovascular Medical and Research Center, Iran University of Medical Sciences, Tehran
Iran
Login to access the Email id

Source of Support: None, Conflict of Interest: None


DOI: 10.4103/rcm.rcm_8_20

Get Permissions

  Abstract 


Aim: One of the most crucial and common occupational hazards in different industries is physical fatigue. Fatigue plays a vast role in all industries in terms of health, safety, and productivity and is continually ranked among the top-five health-related risk factors year after year. The current study focuses on a novel method to detect workers' physical fatigue employing heart rate signals. Materials and Methods: First, domain features are extracted from the heart signals utilizing different entropies and statistical tests. Then, K-nearest neighbors algorithm is used to detect the physical fatigue. The experimental results reveal that the proposed method has a good performance to recognize the physical fatigue. Results: The achieved measures of accuracy, sensitivity, and specificity rates are 78.18%, 60.96%, and 82.15%, respectively, discretely for fatigue detection. Discussion: Based on the achieved results, it is conceived that monitoring of heart rate signals is an effective tool to assess the physical fatigue in manufacturing and construction sites since there is a direct relationship between fatigue and heart rate features. The results presented in this article showed that the proposed method would work well as an effective tool for accurate and real-time monitoring of physical fatigue and help to increase workers' safety and minimize accidents. Conclusion: The results presented in this article shows that the proposed method would work well as an effective tool for accurate and real-time monitoring of physical fatigue and helps to increase workers' safety and minimize accidents.

Keywords: Entropy, hear rate, physical fatigue


How to cite this article:
Darbandy MT, Rostamnezhad M, Hussain S, Khosravi A, Nahavandi S, Sani ZA. A new approach to detect the physical fatigue utilizing heart rate signals. Res Cardiovasc Med 2020;9:23-7

How to cite this URL:
Darbandy MT, Rostamnezhad M, Hussain S, Khosravi A, Nahavandi S, Sani ZA. A new approach to detect the physical fatigue utilizing heart rate signals. Res Cardiovasc Med [serial online] 2020 [cited 2020 Dec 4];9:23-7. Available from: https://www.rcvmonline.com/text.asp?2020/9/1/23/283157




  Introduction Top


Construction is one of the largest and most hazardous industrial sectors, globally. It is a priority industry for worker health and safety and has a disproportionately high rate of recorded accidents.[1] Construction had the highest work-related injury or illness in 2017–2018.[2] Each year, there are at least 60,000 fatal accidents (one of every six fatal accidents at work) on construction sites around the world. Around 25%–40% of work-related deaths occur in construction sites in industrialized countries, even though the sector employs only 6%–10% of the workforce.[2] In some countries, it is estimated that 30% of workers in this industry suffer from back pains or other musculoskeletal disorders.[2]

Fatigue plays a vast role in all industries in terms of health, safety, and productivity and is continually ranked among the top-five health-related risk factors year after year.[3],[4] Fatigue in the workplace is a multidimensional construct that diminishes a worker's performance.[5] It is estimated that fatigue costs more than $18 billion per year in lost productivity alone, of which 84% is due to the reduced performance at work.[3],[6] With increasing concerns regarding occupational safety and health, managing excessive physical workloads of workers is critical to prevent workers' fatigue, injuries, errors, or accidents at physically demanding workplaces.[7] Experience has shown that a preventative safety culture is beneficial for workers, employers, and governments alike.[8]

In previous studies, a wide range of subjective and objective methods have been used to assess the workers' fatigue. The subjective methods used in the previous studies include interviews, questionnaires, Swedish Occupational Fatigue Inventory, and psychomotor vigilance test.[9],[10],[11] However, the results of these methods usually suffer from subjective bias and intrusiveness to a worker's tasks.[10],[11],[12] Wearable sensors enable the continuous monitoring of a wide range of vital signals which can provide early warning systems for workers with high-risk health issues.[5],[13],[14] Wearable sensors are currently being used to manage fatigue in professional athletics, transportation, and mining industries.[5] However, wearable sensor application for fatigue assessment in the industry is on its first steps.

The previous studies harnessed the physiological signals to estimate the fatigue level since it is resulted from physical overexertion and is associated with physiological symptoms. Chang et al.[15] used heart rate to compare emotional stress and physiological strain for different occupations. They investigated the effect of occupation on fatigue and physical symptoms that high-elevation construction workers experience. Venugopal et al.,[16] used multiple time window features to recognize fatigue from surface electromyography signals. She, et al.[17] designed a monitoring and estimation system for the degree of fatigue using heart rate and body movement data. Maman et al.[5] estimated physical fatigue using body move acceleration and heart rate data. Aryal et al.[18] conducted a simulated construction task for monitoring physical fatigue by measuring changes in heart rate, skin temperature, and brain signals. Surangsrirat et al.[19] studied the effect of fatigue on heart rate, body temperature, and skin humidity of people working in high-temperature environment. Zhang et al.[20] used jerk, the time derivative of acceleration, to assess fatigue in physically demanding tasks.

Objective

Based on the conducted literature review, heart rate is the most widely used physiological symptom for monitoring and detecting a worker's fatigue.[21],[22],[23] Monitoring of heart rate signals is an effective tool to assess the physiological strain of subjects in manufacturing and construction sites.[21] It has been proved that there is a direct relationship between physical fatigue and heart rate metrics such as heart rate, heart rate variability, and percentage of heart rate reserve according to the previous studies.[18],[21],[24],[25],[26]

This article proposes a new approach for detecting physical fatigue using heart rate signal monitoring which fills the gap in the literature. In the proposed method, the patterns of physiological signals are studied for fatigue detection. Different entropies and statistical tests are used to extract features.

Using the proposed method, physical fatigue is detected real time using heart signals. Moreover, physical fatigue is predicted more accurately. The proposed method provides an efficient tool to enhance the workers' health and safety in real manufacturing and construction sites and prevent accidents.


  Materials and Methods Top


In this section, the method used for data collection is first introduced. The physically fatiguing task, different methods for entropy calculation, different methods of statistical tests, and classification method are introduced afterward.

Data collection (participants)

The physiological data collected by Sedighi Maman et al.[27] were used to detect the workers' physical fatigue. Their protocol consisted of three physically demanding tasks.[27] Five males and three females from the local community were engaged for a duration of 3.5 months in their research. The age range of the participants was between 18 and 62 years. Two of eight participants were from manufacturing industries and the rest were students exposed to different physical activities.

Physically fatiguing task

In Sedighi Maman et al's study,[27] participants underwent one experimental session. In this session, one physically fatiguing task should be performed within 3 h. The task was divided into 1-h periods. The physical fatiguing task was named as manual material handling (MMH). The MMH task included selection of the packages with different weights, namely 26 kg, 18 kg, or 10 kg, and transferring them to a two-wheeled trolley and moving it to another section and stacking them at the certain locations. The palletization of the package was done in accordance with the packing orders received. One min was the median time for moving one package in one cycle. During 3-h period, each scenario consisted of moving 18 packages summing up to a total of 108 packages as a whole [Figure 1].
Figure 1: An example of manual material handling task[27]

Click here to view


Method

Using statistical tests and fractal dimensions (FDs), some features were extracted in the proposed method. By applying k-nearest neighbors (KNNs) algorithm, we tried to do classification. The best value of K was also selected by changing this parameter from 1 to 40. According to the accuracy of results, when K was 15, we could achieve the best performance. KNN is used because of its classification speed. Actually, there is no learning time. KNN is a lazy learner as it does not learn a classification function from the training data. It remembers the training dataset and uses them for classification.

Nonlinear features based on fractal dimensions

In addition, to mean, standard deviation, minimum, and maximum, some other features such as Higuchi FD (HFD) and Katz FD were also used.

Higuchi fractal dimension

Higuchi introduced the FD calculation of a curve in a plane in 1988.[28] It is nonlinear measure in the time domain for waveform complexity. Time sequences x (1), x (2),…. x (N) can be utilized as signals or discretized functions.[29] A self-similar new time series Ximcan be measured from the starting time sequence as:

Xim: X (m), x (m + k), x (m + 2k),…. x (m + int [(N − k)/k] k) (1)

Where k is the time interval, m = 1, 2,.…., k where m is the initial time, k = 1,…., kmax, int (r) is the integer part of the real number r, kmax is a free parameter. The curves Xim or each of the k time series determines the length of the curve Lm(k).





Where (N − 1)/(int [(Nm)/k] k) is a normalization factor and N is the length of the original time series. The mean value of L (k) was averaged for all m results in Lm(k) for k = 1,……., kmaxas expressed below:



HFD was calculated as the slope of least square linear best fit and an array of mean values is termed as L (k). Hence, HFD is termed as:

HFD = ln (L [k])/ln (1/k) (4)

Windows are the original signal or curve divided into smaller parts with or without overlapping in real scenario. Hence, with or without overlap, HFD values can be estimated.

Katz fractal dimension

The calculation of FD was devised by Katz in 1988.[30] It is a measure of the ratio of curve length L, which is summed up by the Euclidean distances between two consecutive points compared to the maximum distance d from the first point to any point in the frame.[31] It can be termed as the ratio of the length of the curve divided by the line having maximum Euclidean distance from the initial point. The FD is defined as:



Where L is the summation of the Euclidean distances between consecutive points or the total longitude of the curve.



Where d is the planar extension of the curve, which is the distance between the furthest point and the first point in the sequence. D can be termed as:

d = Max (dis [s1, si], i = 1,………., N) (7)

Katz suggested the mean distance between the consecutive points, hence normalizing L and d. Here a = L/N, where N is the step numbers in the curve. The equation (5) becomes:



K-nearest neighbors

The KNN algorithm is a simple, supervised machine-learning classifier that can be used to solve both regression and classification problems. One of the advantages of this algorithm is that it is easy to implement and understand. KNN supposes that similar things exist in close proximity. In other words, similar things are near to each other. KNN categorizes the unknown labels based on similarity measures.


  Results Top


KNN algorithms employed to obtain the accuracy of fatigue detection in MMH task in the present study. The accuracy, sensitivity, and specificity of algorithms were used for their performance comparison.[32],[33],[34],[35],[36],[37] The implementation of these algorithms is represented in [Table 1]. As shown in [Table 1], the accuracy of algorithms will be increased as the value of K is enhanced (goes up). In addition, the value of specificity is higher than the values of sensitivity in KNN algorithms. It is worth to note that the value of AUC will be decreased when the accuracy of algorithm is increased by increasing the values of K.
Table 1: Performance (%) of k-nearest neighbors with different Ks

Click here to view


To draw a meaningful comparison between different KNNs in various conditions, the receiver operating characteristic (ROC) curve of algorithms is shown [Figure 2].
Figure 2: Receiver operating characteristic diagram of k-nearest neighbor algorithm with different Ks

Click here to view



  Discussion Top


Data source used in the present study was collected earlier by Sedighi Maman et al.[27] to detect workers' physical fatigue. They employed five different sensors' features in their study. As a contrast, this research has only dealt with the data extracted from the heart rate sensor. Although the focus of Sedighi Maman et al.[27] research was on using statistical tests, in the current study, some important features were extracted from heart signals using entropies and statistical tests to detect the workers' physical fatigue. KNN algorithms were used for classification. To see the impact of different values of K on the output of algorithms, the value of K started to increase from one. It was concluded that the accuracy of algorithm is increased as the value of K is enhanced. Based on the achieved results, it is conceived that monitoring of heart rate signals is an effective tool to assess physical fatigue in manufacturing and construction sites since there is a direct relationship between fatigue and heart rate features. The results presented in this article showed that the proposed method would work well as an effective tool for accurate and real-time monitoring of physical fatigue and help to increase workers' safety and minimize accidents.


  Conclusion Top


The proposed method can be a practical tool to develop warning systems against high levels of physical fatigue and give an increased amount of rest between tasks to improve workers' safety. Collecting more data sources can be led to more accurate results. As features play an important role in the implementation of algorithm, extracting new features employing other entropies may help to improve the accuracy of the proposed method. It can be concluded that deep-learning algorithms make more meaningful and valuable contributions to the research in this area.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.



 
  References Top

1.
World Statistic-ILO. Available from: https://www.ilo.org ' moscow ' areas-of-work ' WCMS_249278' lang-en. [Last accessed on 2019 Oct 20].  Back to cited text no. 1
    
2.
International Labor Organization, Facts of Safety at Work. Available from: https://www.ilo.org/wcmsp5/groups/public/-dgreports/-dcomm/documents/publication/wcms_067574.pdf. [Last accessed on 2019 Oct 20].  Back to cited text no. 2
    
3.
Shortz AE, Mehta RK, Peres SC, Benden ME, Zheng Q. Development of the fatigue risk assessment and management in high-risk environments (FRAME) survey: A participatory approach. Int J Environ Res Public Health 2019;16:522.  Back to cited text no. 3
    
4.
Lerman SE, Eskin E, Flower DJ, George EC, Gerson B, Hartenbaum N, et al. Fatigue risk management in the workplace. J Occup Environ Med 2012;54:231-58.  Back to cited text no. 4
    
5.
Maman ZS, Yazdi MA, Cavuoto LA, Megahed FM. A data-driven approach to modeling physical fatigue in the workplace using wearable sensors, Appl Ergonomics 2017;65:515-29.  Back to cited text no. 5
    
6.
Caruso CC. Negative impacts of shiftwork and long work hours, Rehabilitation Nursing 2014;39:16-25.  Back to cited text no. 6
    
7.
Hwang S, Seo J, Jebelli H, Lee S. Feasibility analysis of heart rate monitoring of construction workers using a photoplethysmography (PPG) sensor embedded in a wristband-type activity tracker. Automation Construction 2016;71:372-81.  Back to cited text no. 7
    
8.
International Labor Organization. Geneva: Facts on Safety at Work; 2005.  Back to cited text no. 8
    
9.
Chan M. Fatigue: The most critical accident risk in oil and gas construction. Construction Manag Economics 2011;29:341-53.  Back to cited text no. 9
    
10.
Fang D, Jiang Z, Zhang M, Wang H. An experimental method to study the effect of fatigue on construction workers' safety performance. Safety Sci 2015;73:80-91.  Back to cited text no. 10
    
11.
Techera U, Hallowell M, Littlejohn R, Rajendran S. Measuring and predicting fatigue in construction: Empirical field study. J Construction Eng Manag 2018;144:04018062.  Back to cited text no. 11
    
12.
Ahn CR, Lee S, Sun C, Jebelli H, Yang K, Choi B. Wearable sensing technology applications in construction safety and health. J Construction Eng Manag 2019;145:03119007.  Back to cited text no. 12
    
13.
Ananthanarayan S, Siek KA, Health Sense: A Gedanken Experiment on Persuasive Wearable Technology for Health Awareness, Proceedings of the 1st ACM International Health Informatics Symposium. ACM; 2010. p. 400-4.  Back to cited text no. 13
    
14.
Bonato P, Advances in wearable technology for rehabilitation. Stud Health Technol Inform 2009;145:145-59.  Back to cited text no. 14
    
15.
Chang FL, Sun YM, Chuang KH, Hsu DJ. Work fatigue and physiological symptoms in different occupations of high-elevation construction workers. Appl Ergonom 2009;40:591-6.  Back to cited text no. 15
    
16.
Venugopal G, Navaneethakrishna M, Ramakrishnan S. Extraction and analysis of multiple time window features associated with muscle fatigue conditions using sEMG signals. Exp Syst Appl 2014;41:2652-9.  Back to cited text no. 16
    
17.
She J, Nakamura H, Imani J, Ohyama Y, Hashimoto H, Wu M. Verification of Relationship Between Heart Rate and Body Movement for Fatigue Estimation 2014 12th IEEE International Conference on Industrial Informatics (INDIN), IEEE; 2014. p. 775-9.  Back to cited text no. 17
    
18.
Aryal A, Ghahramani A, Becerik-Gerber B. Monitoring fatigue in construction workers using physiological measurements. Automation Construct 2017;82:154-65.  Back to cited text no. 18
    
19.
Surangsrirat D, Dumnin S, Samphanyuth S. Heart Rate, Skin Temperature and Skin Humidity and their Relationship to Accumulated Fatigue. 2019 3rd International Conference on Bio-engineering for Smart Technologies (BioSMART), IEEE; 2019. p. 1-4.  Back to cited text no. 19
    
20.
Zhang L, Diraneyya MM, Ryu J, Haas CT, Abdel-Rahman EM. Jerk as an indicator of physical exertion and fatigue. Automation Construct 2019;104:120-8.  Back to cited text no. 20
    
21.
Gatti UC, Migliaccio GC, Bogus SM, Schneider S. An exploratory study of the relationship between construction workforce physical strain and task level productivity. Construction Manag Econom 2014;32:548-64.  Back to cited text no. 21
    
22.
Mital A, Foononi-Fard H, Brown ML. Physical fatigue in high and very high frequency manual materials handling: Perceived exertion and physiological indicators. Hum Factors 1994;36:219-31.  Back to cited text no. 22
    
23.
Wong DP, Chung JW, Chan AP, Wong FK, Yi W. Comparing the physiological and perceptual responses of construction workers (bar benders and bar fixers) in a hot environment. Appl Ergonom 2014;45:1705-11.  Back to cited text no. 23
    
24.
Hwang S, Lee S. Wristband-type wearable health devices to measure construction workers' physical demands. Automat Construct 2017;83:330-40.  Back to cited text no. 24
    
25.
Jebelli H, Choi B, Lee S. Application of wearable biosensors to construction sites. II: Assessing workers' physical demand. J Construct Eng Manag 2019;145:04019080.  Back to cited text no. 25
    
26.
Lee W, Migliaccio GC. Physiological cost of concrete construction activities. Construct Innovat 2016;16:281-306.  Back to cited text no. 26
    
27.
Sedighi Maman Z, Alamdar Yazdi MA, Cavuoto LA, Megahed FM. A data-driven approach to modeling physical fatigue in the workplace using wearable sensors. Appl Ergonom 2017;65:515-29.  Back to cited text no. 27
    
28.
Higuchi T. Approach to an irregular time series on the basis of the fractal theory. Physica D Nonlinear Phenomena 1988;31:277-83.  Back to cited text no. 28
    
29.
Kesić S, Spasić SZ. Application of Higuchi's fractal dimension from basic to clinical neurophysiology: A review. Comput Methods Programs Biomed 2016;133:55-70.  Back to cited text no. 29
    
30.
Katz MJ. Fractals and the analysis of waveforms. Comput Biol Med 1988;18:145-6.  Back to cited text no. 30
    
31.
Fernández Fraga S, Rangel Mondragón JJ. Comparison of Higuchi, Katz and Multiresolution Box-Counting Fractal Dimension Algorithms for EEG Waveform Signals Based on Event-Related Potentials; 2017. p. 73-83.  Back to cited text no. 31
    
32.
Alizadehsani R, Abdar M, Roshanzamir M, Khosravi A, Kebria PM, Khozeimeh F, et al. Machine learning-based coronary artery disease diagnosis: A comprehensive review. Comput Biol Med 2019;111:103346.  Back to cited text no. 32
    
33.
Alizadehsani R, Roshanzamir M, Abdar M, Beykikhoshk A, Khosravi A, Panahiazar M, et al. A database for using machine learning and data mining techniques for coronary artery disease diagnosis. Sci Data 2019;6:227.  Back to cited text no. 33
    
34.
Alizadehsani R, Habibi J, Alizadeh Sani Z, Mashayekhi H, Boghrati R, Ghandeharioun A, et al. Diagnosing coronary artery disease via data mining algorithms by considering laboratory and echocardiography features. Res Cardiovasc Med 2013;2:133-9.  Back to cited text no. 34
  [Full text]  
35.
Alizadehsani R, Roshanzamir M, Abdar M, Beykikhoshk A, Zangooei MH, Khosravi A, et al. Model Uncertainty Quantification for Diagnosis of Each Main Coronary Artery Stenosis. Soft Computing; 2019.  Back to cited text no. 35
    
36.
Alizadehsani R, Habibi J, Hosseini MJ, Boghrati R, Ghandeharioun A, Bahadorian B, et al. Diagnosis of coronary artery disease using data mining techniques based on symptoms and ecg features. Europ J Sci Res 2012;82:542-53.  Back to cited text no. 36
    
37.
Alizadehsani R, Hosseini MJ, Boghrati R, Ghandeharioun A, Khozeimeh F, Alizadeh Sani Z. Exerting cost-sensitive and feature creation algorithms for coronary artery disease diagnosis. Int J Knowled Dis Bioinform 2012;3:59-79.  Back to cited text no. 37
    


    Figures

  [Figure 1], [Figure 2]
 
 
    Tables

  [Table 1]



 

Top
 
  Search
 
    Similar in PUBMED
   Search Pubmed for
   Search in Google Scholar for
 Related articles
    Access Statistics
    Email Alert *
    Add to My List *
* Registration required (free)  

 
  In this article
Abstract
Introduction
Materials and Me...
Results
Discussion
Conclusion
References
Article Figures
Article Tables

 Article Access Statistics
    Viewed572    
    Printed44    
    Emailed0    
    PDF Downloaded98    
    Comments [Add]    

Recommend this journal