See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/357860413

A feature-based processing framework for real-time implementation of

muscle fatigue measurement

Article in Cluster Computing · January 2022

DOI: 10.1007/s10586-021-03437-7

CITATIONS READS

3 552

3 authors:

Pedro Gonzalez Zamora Victor Hugo Benitez Baltazar

Universidad de Sonora (Unison) Universidad de Sonora (Unison)
221 PUBLICATIONS 23,137 CITATIONS 45 PUBLICATIONS 301 CITATIONS

SEE PROFILE SEE PROFILE

Jesus Pacheco
Universidad de Sonora (Unison)
31 PUBLICATIONS 713 CITATIONS

SEE PROFILE

All content following this page was uploaded by Victor Hugo Benitez Baltazar on 20 May 2022.

The user has requested enhancement of the downloaded file.

Cluster Computing
https://doi.org/10.1007/s10586-021-03437-7 (0123456789().,-volV)(0123456789().
,- volV)

A feature-based processing framework for real-time implementation

of muscle fatigue measurement
P. González-Zamora1 • Victor H. Benitez1 • Jesus Pacheco1

Received: 12 December 2020 / Revised: 29 July 2021 / Accepted: 25 September 2021

Ó The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2022

Abstract
Electromyographic signals (EMGs) are becoming important as a tool for muscle fatigue monitoring. EMGs measure the
electric currents produced in muscle contractions providing information that can be analyzed and processed to evaluate the
conditions of muscles. In this work, we proposed a real-time system that measures muscle fatigue levels based on
Electromyographic signals. We used the Mean Frequency and the Power Spectral Density as features for muscle fatigue
determination. A linear regression model determines the levels of muscle fatigue. Moreover, the system is composed of
EMG wireless sensors allowing it to be used in common activities in the manufacturing industry as manual handling loads.

Keywords EMG signals  Biomedical signals  Occupational medicine

1 Introduction produced during muscle contractions, enabling the study of

physiological muscle properties. The electric currents are
Most of the common injuries in the manufacturing industry known as muscle fiber potentials, and EMGs are composed
are back injuries, which are related to the excess in the of all muscle fiber potentials. As the above is related to the
physical capacities of the workers [1]. Besides, studies neuromuscular activity, EMGs also provide information to
[1, 2] reveal that these injuries can be prevented if an study neuromuscular disorders [11]. EMGs are becoming
evaluation of the risk in working environments is carried important in many fields because of their multiple appli-
out. Disciplines like ergonomics and occupational health, cations as human-machine interfaces, biomedical-medicine
study and evaluate the risk in job activities to produce safer [12]; sports [13]; occupational medicine [14], etc. In
working environments. One crucial aspect to take into human-machine interfaces, EMG can be used to control a
account is muscle fatigue since it is associated with a computer, to drive a car, or to control a robotic arm, among
higher risk of developing musculoskeletal disorders [3]. others [15]. In the sports field, EMG signals are used to
Muscle fatigue can be determined using a variety of measure the muscle activity of athletes. For example, in
methods [4], as measuring the time during which an indi- [16] is described a method to design a pilates training
vidual is able to perform a task with an ergonometric program based on a study of muscle response during
imposed load, keeping a posture during a certain time exercise. In biomedical-medicine, EMG signals are studied
interval, or by measuring the concentration of lactic acid in to find patterns that characterize the signals corresponding
muscle [5]. Moreover, various studies [6–10] have also to each hand-finger and in this way to control powerful
shown that Electromyographic signals (EMGs) provide a prosthesis of upper and lower amputee limbs as it is
method that allows continuous monitoring of muscle fati- described in [17].
gue. This is due to EMGs measure the electric currents EMG signals are generally recorded by non-invasive
sensors placed on the surface of the skin over the muscle.
Then, the recorded signals are studied through spectral
& Jesus Pacheco analysis in order to extract information related to muscle
[email protected]
fatigue. Considering the existing wireless technology of the
1
Department of Industrial Engineering, Universidad de non-invasive sensors combined with an analysis of features
Sonora, Luis Encinas y Rosales S/N, Col. Centro, in real-time, is possible to develop a system to monitor
Hermosillo C.P. 83000, Sonora, México

123
 Cluster Computing

muscle fatigue in real-time. Moreover, the wireless sensors X

M DF
1 X M XM
allow the use of this system in working environments as the pj ¼ pj ¼ pj ð2Þ
j¼1
2 j¼MDF j¼1
manufacturing industry, where the activities performed by
the workers usually require total freedom of arms and legs.
In this paper, a system to monitor the level of muscle
fatigue in real-time using EMG signals is developed. The 3 Materials
system uses non-invasive wireless sensors. An application
was implemented in python for muscle fatigue detection Figure 1 shows the schema of the developed system for
using a linear regression model for fatigue determination. monitoring muscle fatigue in real-time. The proposed
Moreover, the system was evaluated in repetitive tasks method comprises EMG wireless sensors, a wireless base
similar to the manufacturing industry. station, a server, and a client. The function of each element
This paper is organized as follows: in Sect. 2, a is described as follows: the system takes the EMG signals
description of the selected features is given; materials and using the EMG wireless sensors placed over the skin of the
experiment protocol are described in Sect. 3; features volunteer. Then, the EMG signals are transmitted to a
analysis is described in Sect. 4; the monitoring system is wireless base station connected to a local EMG server
described in Sect. 5; experiment results are presented in through a serial cable. The local server sends the data to a
Sect. 6, and Sect. 7 is dedicated to the discussion. client through a TCP connection. Subsequently, the data is
analyzed in real-time by the client.
The EMG wireless sensors and the wireless base station
2 Theoretical background are integrated into the commercial Trigno Wireless System
[23] (see Fig. 2). The Trigno system is composed of eight
The EMGs are non-stationary signals [9]. Their values non-invasive electromyographic sensors that measure the
range go from micro to millivolts; therefore, they are electric currents of muscles. The body size of each Trigno
usually studied in short-time windows where the properties sensor is 27x37x13 mm3 , and the mass is 14 g. The mea-
of EMGs behave as quasi-stationary. In other words, the sured EMG signals obtained by the Trigno sensors are
EMGs data are segmented into short-time intervals for their transmitted to a receiving base machine station via a
analysis. Then, frequency domain features are extracted wireless protocol, which is optimized to minimize data
from the data to obtain information about the muscle latency across sensors. The wireless sensors’ coverage
condition. The above method is widely used in the litera- range is 20 m, allowing taking data without being close to
ture, as is described in various studies [10, 18, 19]. the volunteer.
Moreover, these studies have reported a shift from higher For real-time analysis of the data, a locally based
to lower frequencies in the EMG signals as a manifestation machine is configured as a client. The device used for the
of muscle fatigue [9, 10]. This is due to a decrease in the data analysis is an NVIDIA Jetson Nano [24] (see Figs. 3,
muscle fiber conduction velocity (CV) when one person 4), which is a small computer that allows analyzing
gets tired, affecting the myoelectric signal [20]. This biomedical signals in real-time due to its low latency and
reduction in the frequency can be observed in the Mean less impact on overhead.
Frequency (MNF), and the Median Frequency (MDF) The Jetson Nano was preconfigured with the operating
parameters that are typically used to study muscle fatigue system NVIDIA L4T based on Ubuntu 18.4, which pro-
[21]. The MNF and MDF parameters are obtained from the vides the drivers, kernels, etc., needed for running the
EMG signals’ Power Spectral Density (PSD) [9]. machine. The L4T is a light operative system that allows
The PSD is computed using the Fast Fourier Transform run applications that require high computing time
[22] . The MDF (see Eq. 1) can be described as the fre- performance.
quency at which the PSD of EMG signals is divided into
two regions with equal amplitude, while MNF ( see Eq. 2) 3.1 Experiment protocol
is described as half of the PSD of EMG signals.
X
M X
M According to the ‘‘Manual Handling at work: A brief
MNF ¼ fj pj= pj ð1Þ guide’’ [26] by the Health and Safety Executive (HSE)
j¼1 j¼1
agency of the UK, the weights recommended for manual
where fj is the frequency of the Power Spectral Density handling of loads depend on the height. The maximum
measured at the bin j and pj is the PSD measured at the bin theoretical weight that one worker can handle without
j, and M refers to the length of the frequency bin. being exposed to an injury is shown in Fig. 5 for different
body heights. The safest one is just below the elbow level.

123
Cluster Computing

Fig. 1 Architecture of the

developed system for muscle
fatigue monitoring in real-time

Fig. 2 Trigno wireless system [23] for EMG signals collection. The
wireless sensors allow taking data in working environments like
manual handling loads without affecting the activity of workers

Fig. 4 Illustration of the deltoid muscle. The sensor was placed in red
zone. The figure was taken from [25]

The positions adopted by the volunteer are shown in

Figs. 7 and 8. The volunteer holds the box, which is filled
with books, and then he turns left and puts the box on the
top level of the next desk. Then, immediately, the volunteer
takes the box from the top-level desk, turns right, and puts
the box on the top level of the lower desk. Thus, in this
exercise, the volunteer has two positions: (1) holding the
box just below elbow height and (2) holding the box just
above eye height.
Several muscles can be monitored when performing the
Fig. 3 NVIDIA Jetson Nano [24] used for the analysis of the EMG activity described above. However, in this study, it was
signals in real-time
found that the effort was mainly concentrated in the mus-
cles of the shoulder, more specifically, in the deltoid
The chosen dynamic to test the system consists of
muscle shown in Fig. 4. This is in agreement with studies
handling the box shown in Fig. 6 and transporting it from
reported in [27]. Accordingly, one sensor in the deltoid
one desk to another, taller desk. The height of the smaller
muscle of the right shoulder was placed on each volunteer.
desk is 93 cm, while the height of the taller desk is 150 cm.
As a cross-check, a sensor was also placed in the deltoid
The procedure consists of that the volunteer performs the
muscle of the left shoulder.
activity until the developed system emits an alert when the
In this work, three male students, ages 21, 22, and 21,
volunteer starts reaching the critical point of muscle fati-
participated in this study as volunteers. The volunteer’s
gue. The critical point in this work is when the volunteer
heights are 1.65 cm, 1.72 cm, and 1.74 cm and weigh 56
feels so exhausted.

123
 Cluster Computing

Fig. 5 Recommended weight for manual lifting. The information

shown in this figure was taken from [26]
Fig. 7 Position adopted by the volunteer at elbow height. The height
of the desk is 93 cm

established after doing an offline analysis of PSD and

MNF, and it is described below.
For the offline analysis, all the volunteers performed the
activity described in Sect. 3.1 two times; each lasted 3
minutes. The weight box was varied each time as follows:
1. Without weight: the box was empty. The box’s total
weight was 100 g.
2. With weight: the box was filled with books. The box’s
total weight was 7 kg, which is 40 % above the
recommended weight for eye height.
The PSD was computed for all the volunteers. The PSD for
Fig. 6 Box used for the testing activity. The box size is 50x30x13
cm3
the first volunteer shows a shift from higher to lower fre-
quencies when he performs the activity with weight(see
kg, 60 kg, and 98 kg, respectively. All the participants Fig. 9). In the activity with the box weight of 7 kg, a
signed an Informed Consent form indicating they agreed to narrower peak is observed, which means that the muscle
take part in this study. gets fatigued more quickly when performing the heaviest
task. The same behavior was observed for the second
volunteer (see Fig. 10). The third volunteer does not show
4 Feacture analysis similar results (Fig. 11). However, this can be due to his
physical condition.
Features are computed and analyzed in real-time to deter- The mean frequency (MNF) was also computed for the
mine the critical point of muscle fatigue. The criteria for three volunteers when they were doing the two activities.
the critical point of muscle fatigue determination were In this case, the manifestation of muscle fatigue was more

123
Cluster Computing

Fig. 10 Power Spectral Density computed from EMG signals of the

second volunteer. The same behavior as the first volunteer one is
observed here. There is a more significant peak at lower frequencies
when the volunteer performs the activity with the box of 7 kg. The
dotted line corresponds to the activity with the empty box

Fig. 8 Position adopted by the volunteer at eye height. The height of

the desk is 1.50 cm

Fig. 11 Power Spectral Density computed from EMG signals of the

third volunteer. In this case, the activity with weight show a wider
peak than the other two volunteers

exercise without weight decreases slightly with time. On

the other hand, when volunteer 1 performed the activity
with the box of 7 kg, the reduction of the frequency is
faster.
Fig. 9 Power Spectral Density computed from EMG signals of the
first volunteer. As one can see, there is a more significant peak at 4.1 Linear regression
lower frequencies when the volunteer performs the activity with the
box of 7 kg. The dotted line corresponds to the activity with the empty
box The MNF versus time plots are shown in Figs. 12 and 13.
The graphs show a dispersion of MNF over time that
visible with MNF (see Figs. 12 and 13) than the one with makes it difficult to see the decrement of the frequency,
the PSD. Besides, the MNF shows a clear difference in and consequently, difficult to monitor the muscle fatigue.
both activities. The MNF of volunteer 1 performing the For this reason, the variation of MNF in time was studied

123
 Cluster Computing

The data fit obtained with the linear regression model is

shown in red in Figs. 12 and 13. The fit shows a clear
downward trend of the frequency in Fig. 13, and it is still
possible to see the slight reduction of the frequency as a
function of time in Fig. 12.

5 Monitoring system

Figure 14 shows the flow of the monitoring system. The

EMG signals are processed in the Jeton Nano Machine.
The PSD and MNF features are computed every 0.255
seconds. For each iteration, the system compares the cur-
rent MNF to a baseline to measure the reduction of the
frequency over time. The lower frequency, the higher the
muscle fatigue. The criteria for the determination of muscle
Fig. 12 Mean Frequency (MNF) obtained from volunteer 1 when he
was performing the activity with the 5 kg box. The red dotted line
corresponds to the linear fit using a regression model

Fig. 13 Mean Frequency (MNF) obtained from volunteer 1 when he

was performing the activity with the 7 kg box. The red dotted line
corresponds to the linear fit using a regression model

using a linear regression model. This model assumes a

linear relationship between the MNF and time. The slope
of the linear regression was used to estimate fatigue as it is
described in [28–30]. The equation of the linear regression
is defined as:
y ¼ b 0 þ b1 X þ  ð3Þ
where X is the independent variable, y is the dependent
variable, b0 and b1 are the regression coefficients and  is
the residual error. The linear regression method can be
Fig. 14 Schema of the data acquisition and signal processing to
used to predict the value of the MNFi at timei and in this determine fatigue in a muscle. Trigno wireless sensors and Trigno
way not being affected by the data dispersion. base station belonging to the Trigno Wireless System

123
Cluster Computing

Table 1 Criteria to determine the different intervals of muscle fatigue

Level of fatigue Condition

Yellow MNF \t max

t1 min \ baseline 1

Yellow phase 2 MNF

t1 max \ baseline \t2 max
Critial point MNF
baseline [ t2 max

is presented in Table 1. The system defines three levels of

muscle fatigue: yellow, yellow phase 2, and critical point.
The critical point represents the maximum level of tired-
ness of the volunteer. The values t1 min, t1 max, and t2 max
are the thresholds for each level of fatigue.
The level of muscle fatigue is determined by using the
algorithm 1. As mentioned in 4.1, the MNF and the base-
line used to computed the level of fatigue are obtained
from a linear regression model to be less sensitive to the
data dispersion shown in Figs. 12 and 13. The way to fit the
regression model is shown in Eq. 4.
regressionModel ¼ LinearRegression(time,Y) ð4Þ
where time = {ti ; tiþ1 :::tn } and Y = {yi ; yiþ1 :::yn }, yi is the
MNF computed at ti .
Once the regression model is fitted, the variables are
predicted as follows:
x ¼ regressionModel ðti Þ ð5Þ
The MNF sample collected in the first 30 seconds is fitted
with a linear regression model, and the baseline is obtained
by evaluating the fit at t0 ¼ 0. On the other hand, for the
current MNF, the MNF data sample is fitted in every
iteration. A sliding window, like the one shown in Fig. 15
is used to have the same statistic for the fit and not saturate
the disk memory with data. Moreover, due to First-In,
First-Out (FIFO) principle, the data sample length is 6 Experimental results
always the same. In addition, it is important to mention that
longer-size windows were tested at the beginning of this For testing and validation of the developed real-time sys-
work. The difference in the accuracy of the features tem, the three volunteers described in Sect. 3.1 were
extraction was not significantly different. required again. The test of the system in real-time was
carried out three days after the first data taking described in
Sect. 4, and divided into two days. During the first day, the
volunteers performed the activity with a box of 5 kg twice
to monitor only one of the two (left and right) deltoid
muscles at a time. During the activities, the system emitted
alerts when reaching the three levels of muscle fatigue:
yellow, yellow phase 2, and critical point. Besides, as a
cross-chek, all the volunteers were asked if they were
fatigued when entering the critical point.
Figure 4 shows the results obtained with the 5 kg box for
Fig. 15 Sliding window used for fitting the MNF. For the fit, only the the right deltoid muscle. The time values tell that each
last 30 seconds (green cells) are taking into account. The red cells are
discarded as time goes on volunteer gets fatigued at different times, which can be

123
 Cluster Computing

Table 2 Results obtained with

Volunteer Yellow (min) Yellow phase 2 (min) critical point/feels fatigued
the right deltoid muscle and the
5 kg box during the real-time 1 1.40 1.50 1.55 min/No
testing
2 1.15 1.40 1.45 min/Yes
3 1.20 1.40 2.00 min/Yes
Feels fatigued refers to the answer of the volunteer when entering the critical point

Table 3 Results obtained with

Volunteer Yellow (min) Yellow phase 2 (min) critical point/feels fatigued
the left deltoid muscle and the 5
kg box during the real-time 1 1.30 1.43 1.50 min/Yes
testing
2 1.15 1.35 1.55 min/No
3 1.10 1.20 1.30 min/Yes
Feels fatigued refers to the answer of the volunteer when entering the critical point

Table 4 Results obtained with

Volunteer Yellow (min) Yellow phase 2 (min) critical point/feels fatigued
the right deltoid muscle and the
7 kg box during the real-time 1 0.50 0.54 1.00 min/No
testing
2 0.40 0.45 0.55 min/Yes
3 0.40 0.47 1.53 min/Yes
Feels fatigued refers to the answer of the volunteer when entering the critical point

Table 5 Results obtained with

Volunteer Yellow (min) Yellow phase 2 (min) critical point/feels fatigued
the left deltoid muscle and the 7
kg box during the real-time 1 1.10 1.15 1.22 min/No
testing
2 1.15 1.20 1.30 min/Yes
3 0.42 0.50 1.00 min/Yes
Feels fatigued refers to the answer of the volunteer when entering the critical point

since each of the volunteers has a different physical con- 7 Discussion and future work
dition, as mentioned in Sect. 3.1. On the other hand, the
results obtained when monitoring the left deltoid muscle In this work, a real-time system that monitors muscle
(see Table 2, 3) show a decrease in the volunteers’ per- fatigue using electromyographic signals were developed.
formance. This result can be due to two factors: (1) the The Mean Frequency was used as a feature to determined
volunteers only rested 20 min between the two tests, or (2) muscle fatigue. The system uses a linear regression to
all the volunteers mentioned that due to the activity’s analyse the variation in time of the Mean Frequency. The
movements, they felt more tension in the left shoulder than system used wireless sensors to be more portable and
the right one. defines three levels of muscle fatigue. The activity chosen
The results obtained for the activity with the 7 kg box to validate the system was related to an activity commonly
are shown in Tables 4 and 5 for right and left deltoid performed in the manufacturing industry as the manual
muscle, respectively. The volunteers entering the critical handling loads, which the purpose of develops a tool that
zone more quickly agree with the recommended weight for can be used to measure the risk in this kind of working
manual lifting shown in 5. Moreover, a better agreement environment.
between the real-time system and the feeling of tiredness of
volunteers was observed.

123
Cluster Computing

The obtained results can be summarized as follows: References

– The analyzed data showed that the Mean Frequency
1. Deviani, Triyanti, V.: ‘‘Risk assessment of manual material
could be used as a feature to monitor muscle fatigue as handling activities (case study: PT BRS Standard Industry)’’, IOP
it showed a decrement due to muscle fatigue. Conference Series: Materials Science and Engineering, 277,
– The variation in time of the Mean Frequency was 012043 (2017). https://doi.org/10.1088/1757-899X/277/1/012043
2. Kiani, Fariba, Khodabakhsh, Mohammad, R.: Preventing injuries
studied using a linear regression approach to get less
in workers: the role of management practices in decreasing
sensitive to the data dispersion. injuries reporting. Int. J. Health Policy Manag. 3, 171-177 (2014).
– During the testing phase, the system showed that it took https://doi.org/10.15171/ijhpm.2014.83
longer to get fatigued when the volunteers performed 3. Zenija, R., Henrijs, K., Inara, R.: Measuring muscle fatigue in
relation to the workload of health care workers. Procedia Man-
the activity with the recommended weight by the HSE ufacturing: 6th International Conference on Applied Human
agency. Factors and Ergonomics (AHFE 2015) and the Affiliated Con-
– The system also showed that the volunteers get fatigued ferences, AHFE 2015, 3, 4189-4196 (2015). https://doi.org/10.
more quickly when performing with 43% above the 1016/j.promfg.2015.07.394
4. Cifrek, M., Medved, V., Tonković, S., Ostojić, S.: Surface EMG
recommended weight.
based muscle fatigue evaluation in biomechanics. Clin. Biomech.
– The developed system showed different results when 24, 327–340 (2009). https://doi.org/10.1016/j.clinbiomech.2009.
monitoring the left and right deltoid muscles. The latter 01.010
is interpreted as an effect of the designed activity, 5. Sahlin, K.: Muscle fatigue and lactic acid accumulation. Acta.
Physiol. Scand. Suppl. 556, 83–91 (1986)
which requires more action from the left deltoid
6. Vøllestad, K., Nina.: Measurement of human muscle fatigue.
muscle. On the other hand, the level of muscle fatigue J. Neurosci. Methods. 74, 219-227 (1997). https://doi.org/10.
emitted by the system and each volunteer’s tiredness 1016/S0165-0270(97)02251-6
agrees in 60 %. However, more statistics are needed to 7. Kuthe, Chetan, D.: et al.: Surface electromyography based
method for computing muscle strength and fatigue of biceps
give a more precise performance of the system.
brachii muscle and its clinical implementation. Inform. Med.
Subsequent versions of the system can include an interface Unlocked 12, 34-34 (2018). https://doi.org/10.1016/j.imu.2018.
that sends data of the EMG signals to a cloud system 06.004
8. Cifrek, M., et al.: Surface EMG based muscle fatigue evaluation
allowing remote data analysis. The latter can open a in biomechanics. Clin. Biomech. 24, 327–340 (2009). https://doi.
research line in occupational health where one can monitor org/10.1016/j.clinbiomech.2009.01.010
the muscle fatigue of workers in real-time even if they are 9. Zhang, Z.G., et al.: Time-dependent power spectral density esti-
in another city. mation of surface electromyography during isometric muscle
contraction: methods and comparisons. J. Electromyogr. Kine-
siol. 20, 89–101 (2010). https://doi.org/10.1016/j.jelekin.2008.09.
Acknowledgements Authors thanks to The Mexican Council of Sci-
007
ence and Technology, CONACYT, Mexico, for its support (Grant
10. Karthick, P.A., Ramakrishnan, S.: Muscle fatigue analysis using
236207, CB-2014-01).
surface EMG signals and time-frequency based medium-to-low
band power ratio. Electronics Letters 52, 185–186 (2015). https://
Author contributions All the authors in this paper contribute to the
doi.org/10.1049/el.2015.3460
design and development of the system presented here. The full
11. Raez, M. B., Hussain, M. S., Mohd-Yasin, F.: Techniques of
manuscript was written by P. González-Zamora, and all authors
EMG signal analysis: detection, processing, classification and
contributed with comments on the previous versions.
applications. Biol. Proced. Online 8, 11–35 (2006). https://doi.
org/10.1251/bpo115
Funding This work was supported by the Consejo Nacional de
12. Farfán, F.D., Politti, J.C., Felice, C.J.: Evaluation of EMG pro-
Ciencia y Tecnologı́a de México (Conacyt-México).
cessing techniques using information Theory. BioMed. Eng.
OnLine 9, 72 (2010). https://doi.org/10.1186/1475-925X-9-72
Data availability The datasets generated during and/or analysed dur-
13. Hande, T., Hasan, S.: Surface electromyography in sports and
ing the current study are available from the corresponding author on
exercise. In: Hande Türker, Electrodiagnosis in New Frontiers of
reasonable request.
Clinical Research, pp. 175-194. (2013). https://doi.org/10.5772/
56664
Declarations 14. Marco, G., Babak, A., Roberto, M.: Surface EMG in ergonomics
and occupational medicine. Surf. Electromyogr. Physiol. Eng.
Conflict of interest The authors have no conflicts of interest to declare Appl. 361–391, (2016). https://doi.org/10.1002/9781119082934.
that are relevant to the content of this article. ch13
15. Blanco, A., Catalán, J.M., Dı́ez, J.A., Garcı́a, J.V., Lobato, E.,
Ethical approval All the participants signed an Informed Consent Garcı́a-Aracil, N.: Electromyography assessment of the
form indicating they accept to take part in this study.

123
 Cluster Computing

assistance provided by an upper-limb exoskeleton in maintenance P. González-Zamora was born in

tasks. Sensors (Basel, Switzerland) 19(15), 3391 (2019). https:// San Andres Cholula, Puebla,
doi.org/10.3390/s19153391 Mexico in 1983. He received
16. Steele, C.: Applications of EMG in clinical and sports medicine. the B.Sc degree in computer
IntechOpen (2012). https://doi.org/10.5772/2349 engineering from the Universi-
17. You, K-J., Rhee, K-W., Shin, H-C.: Finger motion decoding dad Autonoma de Puebla, in
using EMG signals corresponding various arm postures 19(1), 2007, and M.Sc and PhD.
54–61 (2010). https://doi.org/10.5607/en.2010.19.1.54 degrees in computer engineer-
18. Knaflitz, M., Bonato, P.: Time-frequency methods applied to ing and Physics from the
muscle fatigue assessment during dynamic contractions. J. Elec- Universidad Complutense de
tromyogr. Kinesiol. 9(5), 337–50 (1999). https://doi.org/10.1016/ Madrid, in 2010 and 2016,
s1050-6411(99)00009-7 respectively. He was a member
19. Masuda, K., Masuda, T., Sadoyama, T., Inaki, M., Katsuta, S.: of the ALICE experiment at the
Changes in surface EMG parameters during static and dynamic LHC CERN from 2007 to 2018.
fatiguing contractions. J. Electromyogr. Kinesiol. 9(1), 39–46 Since 2018, he has been Asso-
(1999). https://doi.org/10.1016/s1050-6411(98)00021-2 ciate Professor in the Industrial Engineering Department of Univer-
20. Merletti, R., Knaflitz, M., De Luca, C.J.: Myoelectric manifes- sidad de Sonora. His research interests include Machine Learning, Big
tations of fatigue in voluntary and electrically elicited contrac- Data, and Grid Computing.
tions. J. Appl. Physiol. 69(5), 1810–20 (1990). https://doi.org/10.
1152/jappl.1990.69.5.1810 Victor H. Benitez received the
21. Angkoon, P., et al.: The usefulness of mean and median fre- B.E. degree inelectronics from
quencies in electromyography analysis. Computational Intelli- the Universidad de Guadalajara,
gence in Electromyography Analysis- A perspective on Current Mexico, in 1999, and the M.Sc.
Applications and Future Challenges, 195-220 (2012). https://doi. and Ph.D. degreesin electrical
org/10.5772/50639 engineering from the
22. Stulen, F.B., DeLuca, C.J.: Frequency parameters of the AdvancedStudies and Research
myelectric signal as a measure of muscle conduction velocity. Center, CINVESTAV-IPN, in
IEE Trans. Biomed. Eng. 28(7), 515–23 (1981). https://doi.org/ 2002 and 2009, respectively. He
10.1109/tbme.1981.324738 has worked inthe manufacturing
23. Trigno Wireless System, https://delsys.com/trigno/. Accessed 7 industry related to the manu-
Dec 2020 facture of printed circuits
24. Jetson Nano NVIDIA, https://developer.nvidia.com/embedded/ boards, and in the areaof testing
jetson-nano-developer-kit. Accessed 7 Dec 2020 engineering. He is currently a
25. Primal Pictures Ltd: Trigger point Deltoid 1. (2020) Full Pro-fessor of mechatronics
26. Health And Safety Executive (HSE), Manual handling at work: a with the Industrial Engi-neering Department, Universidad de Sonora.
brief guide. HSE Books, London (2020) His research interests includemyoelectric control systems, neural
27. Sung-kwang, J., Yoo, W-G: Electromyography activity of the networks applied to control electrome-chanical systems, and
deltoid muscle of the weight-bearing side during shoulder flexion embedded systems applied to the Internet of Thing
in various weight-bearing positions. J. Phys. Ther. Sci. 27(10),
3285–3286 (2015). https://doi.org/10.1589/jpts.27.3285 Jesus Pacheco received the
28. Thongpanja, S., Phinyomark, A., Phukpattaranont, P., Limsakul, bachelor’s degreein electronics
C.: Mean and median frequency of EMG signal to determine engineering with a focus on
muscle force based on time-dependent power spectrum. Elek- digitalsystems and the M.Sc.
tronika Ir Elektrotechnika 19(3), 51–56 (2013). https://doi.org/10. degree in computer sciencefrom
5755/j01.eee.19.3.3697 the Tecnol_ogico de Hermosillo
29. Oliveira, R.S., Pedro, R.E., et al.: Spectral analysis of elec- and the Ph.D. degree from the
tromyographic signal in supramaximal effort in cycle ergometer Electrical and Computer Engi-
using Fourier and Wavelet transforms: a comparative study. neering Department, The
Revista Andaluza de Medicina del Deporte 5(2), 48–52 (2012). University of Arizona. He is
https://doi.org/10.1016/S1888-7546(12)70008-4 currently a Full Professor with
30. Smith, L.H., Kuiken, T.A., Hargrove, L.J.: Evaluation of Linear the IndustrialEngineering
Regression Simultaneous Myoelectric Control Using Intramus- Department, Universidad de
cular EMG. IEEE Trans. Biomed. Eng. 63(4), 737–746 (2016). Sonora,and the Director of the
https://doi.org/10.1109/TBME.2015.2469741 NSF Center for Cloudand
Autonomic Computing. His research interestsinclude cyber security
Publisher’s Note Springer Nature remains neutral with regard to for the Internet of Things and cyber-physical systems
jurisdictional claims in published maps and institutional affiliations.

123

View publication stats