A NEW APPROACH TO DETERMINING THE

ACTIVE POTENTIAL LIMIT OF AN

Abstract— In the realm of electromyography (EMG) signal patient health through potentials from human muscles. A signal
processing, accurately determining the active potential is that measures the electrical potential of the muscles and is the
paramount. This article unveils a groundbreaking approach to main criterion for movement is an electromyography (EMG)
assessing the active potential of an EMG signal, contrasting its signal. This signal is highly complex by nature and is
efficacy with prevalent algorithms. Central to this innovative distinguished from other signals by its extreme sensitivity to the
method is the analysis of the standard deviation of the EMG signal, influence of the external environment. Since it belongs to the
a strategy that facilitates instantaneous high-precision processing. category of weak-frequency signals, its processing is a complex
Preliminary results have been promising, with the method
process [10].
boasting an average error of merely 1.54% and a deviation value
of 18.9. Notably, this approach trumps alternative algorithms in EMG signal classification is currently being used in the areas
terms of precision, speed, and computational simplicity. As the of motion recognition, muscle strength analysis, and
fundamental intermediary speed in signal-driven systems, EMG's neuromuscular disease detection. The general procedure for
active potential assessment is vital. The presented method, with its classifying an EMG signal is given below (Figure 1):
superior performance metrics, holds significant promise for
advancing Human-Machine System research, potentially Data collection: signal recording, collection.
revolutionizing their operational efficiency and accuracy. The
Pre-processing of data: initial processing of collected signal
implications of this research are far-reaching, underscoring the
method's potential as a pivotal tool in future EMG-based studies data
and applications. Parameter extraction: appropriate parameters are obtained
to help to class the data [12].
Keywords— electromyography, biosignals, active limit,
frequency, standard deviation, mean error. Parameter selection and reduction: Select parameters for
the main objective in this and calculate them to reduce the time,
I. INTRODUCTION reducing the amount of information necessary for grading. The
Currently, a huge number of studies and results have been choice of appropriate parameters is a critical stage since
obtained in areas such as processing biosignals [1][2][3], functions should provide a high differentiation between
classification, and early detection of diseases in the world with categories [13][14][15][16].
artificial intelligence [4][5][6], the creation of human-machine Classification model selection: It is necessary to choose the
systems [7]. appropriate classification model based on the data. Classifier
We comprehend the signals that propagate from the human efficiency may also depend on the type of use. functions
body by biosignal. They can be a description of natural [17][18].
physiological phenomena that occur in the human body. Classifier training and testing: The supervisory training is
Biosignals are the main object of Biosystems. carried out on the Mal using the training sample. At the same
Nowadays, there are many types of biosignals as electrical, time, classification models are tied to a specific training sample,
magnetic, optical, and chemical, thermal, mechanical [8][9]. providing a high result in training new data.
These, in turn, are divided into such types as endocrine, nervous,
Pre-processing of Parameter
cardiovascular, visual, auditory, muscular, respiratory, and Data collection
data extraction
gastrointestinal system signals according to the system of origin.
Biosignals are much weaker and have very small values like
them, with very small values of frequency and amplitude [10].
Parameter
When it comes to work on human health rehabilitation of Classification Classifier training
selection and
patients depending on the physiological data of the patient [10], reduction
model selection and testing
diagnosis issues also attract attention in the world [11]. For
example, we can cite scientific research that is being done on Fig. 1. EMG signal classification steps
 II. MOTIVATION The motivation behind advancing EMG signal processing is
The human body is a marvel of engineering, a complex clear. As we stand on the cusp of significant technological
system where various components interact seamlessly to advancements, the need for precise, efficient, and adaptable
perform a myriad of functions. Central to these functions are EMG analysis becomes more pressing. The potential benefits,
biosignals - the silent communicators of the body's internal state. from revolutionizing healthcare to creating intuitive human-
These signals, often subtle and intricate, offer a window into the machine interfaces, are profound. But to realize this future, we
physiological processes at play, from the rhythm of a heartbeat must invest in research, foster interdisciplinary collaborations,
to the electric hum of a neuron firing. In this context, the and above all, remain curious and open to innovation. The
electromyography (EMG) signal stands out due to its crucial role human body's silent communicators, the biosignals, have much
in understanding muscle activity. to tell. It's up to us to listen, decode, and harness their potential.

As we navigate the 21st century, our society witnesses a III. MATERIALS AND METHODS
profound intersection of biology, technology, and artificial The EMG signal parameters given above are widely used in
intelligence. This confluence presents unprecedented biomechanics, rehabilitation, clinical diagnostics, and other
opportunities and challenges. One of the most pressing areas of research based on the methods of Character
challenges is the accurate and efficient processing of biosignals. Engineering. Calculation of the given parameters is considered
EMG, with its intricate patterns, demands advanced techniques important to distinguish from the signal area at the time of
for accurate interpretation. Why? Because these signals hold the muscle contraction. Therefore, it is necessary to determine the
key to numerous applications that can revolutionize healthcare, activation part of the signal in the process of classifying the
rehabilitation, and human-machine interfaces. EMG signal. One way to determine the active muscle limit is
EMG signals are pivotal in diagnosing neuromuscular through visual inspection. Visual inspection can be accurate but
diseases. Conditions like myopathy, neuropathy, and motor complex, while in Real-time several algorithms have been
neuron diseases manifest in distinct EMG patterns. Thus, precise developed that detect automatic active potentials because of the
analysis can facilitate early detection, allowing timely low efficiency of this method.
intervention and better patient outcomes. Furthermore, for Some algorithms automatically detect the active potential
patients undergoing rehabilitation post-injuries or surgeries, part of the EMG signal and a comparative analysis of the
EMG data can provide insights into muscle recovery, guiding proposed algorithm is below.
therapists in devising effective recovery plans.
A. Determination of muscle activity.
As we inch closer to a future dominated by robotics and AI,
the seamless integration of humans and machines becomes Research has been done to determine the starting point of
paramount. EMG signals can be harnessed to create intuitive muscle activation in the EMG signal. Visual inspection of the
human-machine interfaces. Imagine a world where prosthetic signal is carried out using special algorithms. Visual detection
limbs move as naturally as biological ones, driven by the user's cannot be used in real-time. Several algorithms that determine
EMG signals. Or consider advanced exoskeletons that amplify the beginning and end of the muscle's active potential have been
human strength and endurance, all synced to the wearer's muscle cited [19].
activity. The possibilities are endless.
Given the profound implications of EMG analysis, it's B. The single limit algorithm.
evident that traditional methods, while foundational, have The onset of muscle activation is estimated as a point where
limitations. Visual inspection, for instance, is subjective and not the amplitude of the rectified EMG signal exceeds the
feasible for real-time applications. The single-limit and two- predetermined threshold. However, the fact that the operation of
limit algorithms, though more objective, are still constrained by this method depends on the choice of the boundary creates a
predetermined thresholds, leading to potential inaccuracies. tendency to miscalculate.
This underscores the need for innovative algorithms, ones C. The two-limit algorithm
that are adaptable, efficient, and accurate. Algorithms like the This is an improved representation of the above method by
character transformation method, process EMG signals limiting signals with sensitivity. The onset of muscle activation
dynamically, considering their unique characteristics. Such is determined only when a predetermined number exceeds the
approaches promise higher precision, paving the way for second limit through sequential experiments.
applications that were previously deemed challenging.
D. The character transformation algorithm
While advanced algorithms form the backbone of modern
EMG analysis, a holistic approach is essential. This means Reduces the values of a signal by processing it as modular
understanding the nuances of human physiology, the intricacies or square values and analyzes important data on how it makes
of the EMG signal themselves, and the technological signal records towards the X-axis.
constraints. It also fosters interdisciplinary collaborations, Algorithm 1: Character transformation algorithm.
where biologists, engineers, data scientists, and clinicians come
together. Such methods can propel the field forward, ensuring Step 1. Window sizes are inserted to process the
that the algorithms developed are not just technically sound but electromyography signal into pieces.
also clinically relevant.
 Step 2. A threshold of 0.01 is placed on the signal, and at this Step 5. The value of the Signal dispersion is 1<S<the offset
threshold, the signal is passed through the window. interval in which the cycle is executed.
Step 3. The maximum length of the incoming signal is Step 6. When the Signal dispersion is less than the S*k value,
calculated, and the signal passing through the window is the sum of the window size with the onset value goes through
multiplied by 1.5, the modulus of the maximum signal value in the cycle.
the range is determined.
Step 7. If k exceeds the limit that point marks the beginning
Step 4. The maximum values of the signal are determined by of the area of the signal active potential.
dividing each window into pieces with a ratio such as 2, 4, or 6.
The standard deviation of the signal is found as follows:
Step 5. If the maximum value k is less than the limit the cycle
will continue, if the signal value is greater than the specified 1 N
limit, that point will be the active part of the signal. =
N
 ( x − x)
i =0
i ()
A change in sign occurs when the muscle is active. When the
muscle is active, it is observed that the voltage values are quite
where N - is the number of signal values, 𝑥𝑖 - signal value,
different. The initial signal values determine the location of the
𝑥 - signal average. The total length of the EMG signal is
change, and this point is considered to be the muscle active point
calculated, and the values are adjusted to the absolute value. Is
(algorithm 1).
the dispersion of the corrected signal by window size.
The Komi algorithm is a popular algorithm that can
It is important to carry out a comparative analysis of these
determine the beginning limit of the activity of an EMG signal.
algorithms. Determining the exact starting point of the active
Determining the limits of the onset of muscle activity is an
potential and extracting parameters from exactly that point of
important element in the biomechanical analysis of human
the EMG signal has a positive effect on speediness on the one
movement. It was proposed by scientists Komi and Kavanagh in
hand and accuracy on the other when classifying ((Figure 2).
1979. In the algorithm, the limit value was taken to be 30 MV,
since the studies gave it that the voltage produced by the human
muscles was considered the lowest limit of diapason [20]. It is
calculated as follows.
Algorithm 2: Komi algorithm.
Step 1. A threshold of Onset = 0.03 is placed on the signal,
and at this threshold, the signal is passed through the window.
Step 2. The Signal length is calculated and the set point is Fig. 2. Comparative graph of results for determining the active limit of the
compared with the set limit. EMG signal

E. The Teager-Kaiser energy operator (TKEO) algorithm TABLE I. COMPARATIVE ANALYSIS OF ALGORITHMS FOR DETERMINING
THE ACTIVE LIMIT OF THE EMG SIGNAL THROUGH VARIOUS ALGORITHMS
This algorithm mainly performs calculations along signal
wavelengths and is distinguished from others by its extreme Current
Average Average
sensitivity. It is found as follows: Algorithms
error deviation
signal
order
Single limit algorithm 3,1 % 32.2 98
U[i] =  ( xi ) = xi 2 − ( xi +1  xi −1 ) ()
Two-limit algorithm 2,56 % 22.6 156
Proposed algorithm


1,54 % 18.9 352
E  i , if i = 0 or i = N −1 Character transformation
2,48 % 25.8 718
U i  = () algorithm
E i  − ( E i +1* E i −1), otherwise
2
Komi algorithm 2,95 % 32.8 956
Teager-Kaiser energy
2,02 % 27.5 402
operator (TKEO) algorithm
here, 𝑥𝑖 - the i - value of the EMG signal.
F. The proposed algorithm. Table 1. presents the results of these algorithms for
Step 1. Window size k, mean of EMG signal values – 𝜇, determining the active moments of the EMG signal. Analysis
standard deviation of EMG values – 𝜎, h - values such as the shows that the largest mean error was observed in the single
variable setting the threshold level are entered. threshold (Single threshold) algorithm (Table 1). The proposed
algorithm's average error was 1.54%, and the average deviation
Step 2. The limit value tℎ𝑟𝑒𝑠ℎ𝑜𝑙𝑑=𝜇+ℎ𝜎 is entered and the value was 18.9.
signal length M is calculated.
Based on these methods, it can be seen that the following
Step 3. Signal values equal to M in length are adjusted, that accuracies are achieved when the movement classification is
is, those with negative values are discarded. performed using the EMG signal in real time.
Step 4. The adjusted signal is the S dispersion between value
1 and window size.
 [5] R. Nasimov, N. Nasimova and B. Muminov, "Hybrid Method for
Evaluating Feature Importance for Predicting Chronic Heart Diseases,"
2022 International Conference on Information Science and
Communications Technologies (ICISCT), Tashkent, Uzbekistan, 2022,
pp. 1-4, doi: 10.1109/ICISCT55600.2022.10146969.
[6] R. Boostani and M. H. Moradi, Evaluation of the forearm EMG signal
features for the control of a prosthetic hand," Physiological Measurement,
vol. 24, pp. 309-319, 2003.
[7] K. Zohirov, "Classification Of Some Sensitive Motion Of Fingers To
Create Modern Biointerface," 2022 International Conference on
Information Science and Communications Technologies (ICISCT),
Tashkent, Uzbekistan, 2022, pp. 1-4, doi:
10.1109/ICISCT55600.2022.10146846.
[8] H. P. da Silva, A. Fred, and R. Martins, "Biosignals for Everyone," in
Classification of actions was performed based on the IEEE Pervasive Computing, vol. 13, no. 4, pp. 64-71, Oct.-Dec. 2014,
Random forest (RF) algorithm, which belongs to the category of doi: 10.1109/MPRV.2014.61.
machine learning methods. RF algorithm is one of the most [9] Swapna M, Viswanadhula UM, Aluvalu R, Vardharajan V, Kotecha K.
Bio-Signals in Medical Applications and Challenges Using Artificial
suitable methods for real-time classification. The experiment Intelligence. Journal of Sensor and Actuator Networks. 2022; 11(1):17.
showed that our proposed algorithm achieved 99% accuracy in https://doi.org/10.3390/jsan11010017
classification by RF algorithm. [10] C. Scuro, P. F. Sciammarella, F. Lamonaca, R. S. Olivito, and D. L. Carni,
"IoT for structural health monitoring," in IEEE Instrumentation &
IV. CONCLUSION Measurement Magazine, vol. 21, no. 6, pp. 4-14, December 2018, doi:
10.1109/MIM.2018.8573586.
Regarding the studies carried out in this article, we can
[11] Palumbo A, Vizza P, Calabrese B, Ielpo N. Biopotential Signal
conclude: Monitoring Systems in Rehabilitation: A Review. Sensors (Basel). 2021
A new method has been developed to determine the real- Oct 28;21(21):7172. doi: 10.3390/s21217172. PMID: 34770477;
PMCID: PMC8587479.
time active potential limit of the EMG signal. The proposed
[12] A. Feltane. Time-frequency-based methods for non-stationary signal
method's average error was 1.54% and the average deviation analysis with application to EEG signals. A dissertation for the degree of
value was 18.9. A classification accuracy of 99% was achieved. Doctor of Philosophy in Electrical Engineering, University of Rhode
It outperforms other algorithms that determine the active Island 2016. pp. 149. ProQuest Number:10102283.
potential by its accuracy, agility, and ease of computation. THIS [13] Phinyomark, A.; Scheme, E. EMG pattern recognition in the era of big
METHOD IS IMPORTANT IN FUTURE RESEARCH since data and deep learning. Big Data Cogn. Comput. 2018, pp. 2-21.
EMG is the most basic mezzanine speed in signal-driven [14] Omar Abdul-hadi. Machine Learning Applications to Robot Control. A
systems or devices. dissertation for the degree of Doctor of Philosophy in Engineering -
Mechanical Engineering, University of California, Berkeley, 2018. pp.
REFERENCES 99. ProQuest Number:10817183.
[15] A. Phinyomark, P. Phukpattaranont, and C. Limsakul, “Feature reduction
[1] A. Turgunov, K. Zohirov, R. Nasimov and S. Mirzakhalilov, and selection for EMG signal classification", Expert Systems With
"Comparative Analysis of the Results of EMG Signal Classification Applications, vol. 39, pp. 7420-7431, 2012.
Based on Machine Learning Algorithms," 2021 International Conference
on Information Science and Communications Technologies (ICISCT), [16] A. Phinyomark, C. Limsakul, and P. Phukpattaranont, A Novel Feature
Tashkent, Uzbekistan, 2021, pp. 1-4, doi: Extraction for Robust EMG Pattern Recognition," Journal of Computing,
10.1109/ICISCT52966.2021.9670108. vol. 1, no. 1, pp. 71-80, 2009.
[2] A. Turgunov, K. Zohirov, S. Rustamov and B. Muhtorov, "Using [17] B. Hudgins, P. Parker, and R. N. Scott, A New Strategy for Multifunction
Different Features of Signal in EMG Signal Classification," 2020 Myoelectric Control," IEEE Transactions on Biomedical Engineering,
International Conference on Information Science and Communications vol. 40, no. 1, pp. 82-94, 1993.
Technologies (ICISCT), Tashkent, Uzbekistan, 2020, pp. 1-5, doi: [18] H. Fayed and A. Atiya, “A Novel Template Reduction Approach for the
10.1109/ICISCT50599.2020.9351392. Nearest Neighbor Method”, IEEE Trans on Neural Network, Vol. 20, No.
[3] A. Ekelem. Control Methods for Improving Mobility for Persons with 5, May 2009. pp:890
Lower Limb Paralysis. Dissertation for Doctor of Philosophy in [19] Mateusz Magda. EMG onset detection – development and comparison of
Mechanical Engineering September 30, 2018, Nashville, Tennessee. algorithms. Thesis, Wrocław University of Technology, pp. 48. 2015.
ProQuest Number:13877286. pp.156. [20] Solnik, S., Rider, P., Steinweg, K., DeVita, P., and Hortobágyi, T.
[4] Abdusalomov AB, Nasimov R, Nasimova N, Muminov B, Whangbo TK. Teager–Kaiser energy operator signal conditioning improves EMG onset
Evaluating Synthetic Medical Images Using Artificial Intelligence with detection. European journal of applied physiology 110, 3 (2010), 489–
the GAN Algorithm. Sensors (Basel). 2023;23(7):3440. Published 2023 498.
Mar 24. doi:10.3390/s23073440