Volume 7, Issue 12, December – 2022 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165

Physical Exertion Detection

(Using Machine Learning and IOT)
Nandini Sunil Mutha Rutuja Shatrughna Mandhare
Dept. of Information Technology Dept. of Information Technology
Sinhgad College of Engineering Sinhgad College of Engineering
Vadgaon BK, Pune, Maharashtra, India Vadgaon BK, Pune, Maharashtra, India

Devyani Nitin Mahajan(Author) Omkar Rajesh Dukare

Dept. of Information Technology Dept. of Information Technology
Sinhgad College of Engineering Sinhgad College of Engineering
Vadgaon BK, Pune, Maharashtra, India Vadgaon BK, Pune, Maharashtra, India

Prof. Tejal. H. Patil

Dept. of Information Technology
Sinhgad College of Engineering
Vadgaon BK, Pune, Maharashtra, India

Abstract:- With the advancement of technology, we are especially in older adults with a habitualness of 1.3%, 1.5, and
able to use Arduino in this project to digitally sense body 8.4% in the 55–64, 64–74, and 75 and older segments,
temperature and heart rate. Arduino is primarily used respectively. Hospital staff face serious difficulties when faced
because it has the ability to perceive its surroundings by with the task of caring for multiple patients at the same time.
receiving data from various sensors and change it by Problems like waiting in queue, travel schedule, patient
controlling lights, motors and other actuators. Arduino transfer, waiting for doctor etc. are some of the problems faced
programming language is used to program the by the patients. In an emergency, the situation may worsen.
microcontroller of the board. The goal of this work is to 24/7 monitoring of a critical patient is very important to reduce
develop a system that can remotely monitor body the risk of life threatening. In this situation, our app has had a
temperature and heart rate in real time with the big impact on healthcare services. It also reduces hospital
measurement results displayed on a web server. This operating costs. In medical science, this application has several
research involves a number of phases, including the advantages, such as ease of use, reduced risk of infection, and
generation of research hypotheses based on literature increased mobility. Using this system, it is very easy to monitor
reviews and the design of a hardware and software system. several patients at the same time. A physical effort detection
system has been designed that is capable of performing various
Keywords:- Arduino, Sensor, SVM Classifier, Feature types of functions within frame time, accuracy and cost. This
Extraction, Pre-Processing system is cheap and can be controlled remotely. The biological
factors of the patients are detected by a sensor. Using the sensor
I. INTRODUCTION with Arduino made the physical effort detection system more
efficient. Using various machine learning algorithms such as
Nowadays, the cause of unhealthy human lifestyle is wavelet transform, low pass and support vector machine
based on many factors such as irregular eating habits, no diet, (SVM), the input is processed and classified for the desired
environmental pollution, lack of proper exercise, work on output. In this situation, our app has had a big impact on
indefinitely, restlessness and increases stress levels, leading to healthcare services.
greater disaster in human health. In many countries, the
percentage is nearly 40% of children, middle-aged people and II. BACKGROUND
working women are affected by an unhealthy lifestyle. Due to
the busy schedule in our daily life, we rarely have time to focus Advances in medical technology are rapidly changing the
on our health, resulting in various health risks. Also, it would e-health system. Today, advanced medical technology
be difficult for the doctor to monitor the patient continuously. contributes greatly to our lives. This will help improve and
It is also a bit difficult for an individual patient to monitor their save countless lives around the world. Medical technology is
health status and get advice from a doctor. Existing healthcare a broad field where renaming plays an important role in
application systems, such as, patient monitoring systems, improving health. According to the World Health
health management systems are not up to par in providing Organization, the impact of physical fatigue on human health
sufficient information or services to patients. Nowadays, has increased in recent years. “Fatigue” has been described as
people are suffering from various forms of diseases and many a risk for construction accidents in several studies, but little
health issues such as CHF (Chronic Heart Failure) are normal attention has been paid to this risk. Fatigue affects worker
in the aged people. CHF is the responsible for hospitalization

IJISRT22DEC763 www.ijisrt.com 188

Volume 7, Issue 12, December – 2022 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
health and safety and must be measured and managed to Zhaolong Ning; Lei Guo; Bin Hu; Tie Qiu and Ricky Y. K.
prevent serious injuries. Kwok (2021).

III. LITERATURE SURVEY In this article, the authors propose to create a cost-
effective home health monitoring system for IoMT divided into
“E-Healthcare Monitoring System using IoT with two networks. H. Intra-Wireless Body Area Network (WBAN)
Machine Learning Approaches “. Published by IEEE and and Beyond-WBAN. Highlighting the features of IoMT, the
authors are: Brahmaji Godi; Sangeeta Viswanadham; Appala cost to the patient depends on the age of information (AoI),
Srinuvasu Muttipati; Om Prakash Samantray; Sasi Rekha medical importance, and energy consumption. For intra-
Gadiraju (2020). WBAN, cooperative games are formulated to allocate wireless
channel resources. On the other hand, for Beyond-WBAN, a
This paper introduces the IoT application framework of decentralized, non-cooperative game is proposed to minimize
the e-healthcare monitoring system "EHMS". Design advanced the system-wide cost of IoMT, considering individual
automation systems in combination with machine learning rationality and potential selfishness. . They prove that the
(ML) algorithms. This system allows EHMS to connect, proposed algorithm can achieve Nash equilibrium. In addition,
monitor and make decisions for proper diagnostics. The author the time complexity of the algorithm and upper bounds on the
currently recommends an e-health framework system "EHMS". number of patients benefiting from her MEC are theoretically
An application model developed using various machine derived. Performance evaluations demonstrate the
learning models. Patient data is collected using portable IOT effectiveness of the proposed algorithm in terms of overall
sensors. Data collected from myriad wearable health devices is system cost and number of patients benefiting from MEC. The
processed by e-health monitoring systems. EHMS then applies authors aim to minimize the overall system cost of IoMT by
machine learning approaches to the raw data to analyze patient planning transmission and computing resources. In this article,
health for better decision-making and diagnosis. Disability the author analyzed her MEC-enabled 5G home health
Interpretation Reports and Physician Appointments are monitoring of her IoMT within WBAN and he beyond WBAN.
acknowledged by patients, physicians, and caregivers. In this
proposed system, we propose that the e-health monitoring IV. PROPOSED SYSTEM
system uses wearable IoT health devices that transmit real-time
information from the human body. The system aims to use The designed system consists of hardware and software.
Internet of Things (IoT) architecture and machine learning Hardware components include the DTH11 sensor, which is
techniques to develop new application models that provide commonly used temperature and humidity sensor with a
better solutions and significant improvements to many remote dedicated NTC to measure the temperature and 8-bit micro-
healthcare services. controller to output the temperature and humidity values to
serial data. Arduino-Uno is a microcontroller Everything you
“Condition Monitoring Based on Partial Discharge need to support your microcontroller is included; Simply
Diagnostics Using Machine Learning Methods: A connect it to your computer with a USB cable or power it with
comprehensive state-of-the-art review” published by IEEE an AC/DC adapter or battery, and you're ready to go The
Transactions on Dielectrics and Electrical insulation Vol. 27, software acquires the data accepted by the sensor This read data
No. 6 and authors are: Shibu Lu; Hua Chai; Animesh Saaho is preprocessed using a low pass filter A lowpass is a filter that
and B.T. Phung (2020). passes signals with frequencies below a selected cutoff
frequency and attenuates signals with frequencies above the
This white paper comprises a state-of-the-art overview of cutoff frequency We use Wavelet transformation for feature
machine learning (ML)-based intelligent diagnostics applied to extraction The wavelet transform is a mathematical tool for
partial discharge (PD) detection, localization, and pattern signal analysis Waveform analysis is effective when the
recognition. Machine learning techniques, especially those analyzed signal has transitions, discontinuities or distortions in
developed in the last five years, are examined and categorized the voltage and current waves The SVM classifier is used for
as traditional machine learning or deep learning (DL). Key further classification Support Vector Machines or SVMs are
characteristics of each technique, such as input signal type, core one of the most popular learning algorithms for classification
technique, sample rate, and accuracy, are summarized and and regression problems The proposed method will help to
compared in detail. I will briefly discuss the strengths and identify physical exertion. Accurate and timely monitoring of
weaknesses of these different machine learning algorithms. physical exertion can improve worker safety and help prevent
Additionally, technical obstacles have been identified that accidents The proposed method could be used to create
hinder the application of intelligent partial discharge warning systems for excessive physical exertion, as well as to
diagnostics in the industry. For example, create enhanced break times to improve worker safety The
imbalanced/insufficient datasets, data discrepancies, difficulty microcontroller is programmed to receive input from the heart
in cost-effective real-time delivery. Finally, possible solutions rate sensor when a finger is inserted and continuously display
are proposed and future research directions are suggested. the values on a waveform Wave input form to Arduino, then it
takes that input as input for python block The Python block
“Mobile edge computing enabled 5G health monitoring contains data loading, feature extraction (WT), and the SVM
for the internet of medical things: A decentralized game classifier The result will come in the form of body temperature,
theoretic approach” published by IEEE Journal on selected sweat percentage, heart rate and level of physical exertion.
areas in communications Vol. 29, No.2 and authors are:

IJISRT22DEC763 www.ijisrt.com 189

Volume 7, Issue 12, December – 2022 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
A key term in a low-pass circuit is the CUT-OFF
FREQUENCY. The cutoff frequency is the frequency above
which no change in voltage as a function of time may enter the
circuit. For example, if the low pass had a cut-off frequency of
30 Hz, the line voltage type of interference (60 Hz) would be
filtered out, but the 25 Hz signal could pass. Also in a digital
circuit, a low-pass filter can be used to bounce an input from a
momentary contact button pressed by a person, or even to
switch a relay connected to a counter input. The low-pass filter
can be constructed from a resistor R and one capacitor C. The
cut-off frequency Fc is determined by the formula:

B. Wavelet Transform
Fig 1.System Architecture Waveform analysis is very effective When analogy
signals containing transitions, discontinuities or Distortion of
V. ALGORITHMS voltage and current waves Basic concept of wavelets The
analysis consists in choosing an appropriate wavelet function,
A. Low Pass Filter called “mother”. wavelets" then analyzed by moving and
A low pass is a filter that passes signals with a frequency expanding version of this wavelet The main advantage of WT
lower than the selected cutoff frequency and attenuates signals over STFT is that it is used High frequency short window, low
with frequencies higher than the cutoff frequency. The exact frequency long window. This Wavelet transform can be used
frequency response of the filter depends on the design of the for short time intervals with high frequency component and
filter. The filter is sometimes called a high or high cut filter in the long separation of low frequency components, making it.
audio applications. The lower filter is a supplement to the upper The right tool to analyze high frequency transients in their
filter. presence low frequency components The wavelet transform is
intrinsically more suited to non-stationary and non-periodic
broadband signals.

There are two types: Waveform transformations:

continuous and discrete The definition of each type is As
shown in Fig. The main difference between the two types is
that the continuous wavelet transform (CWT) uses a sequence
of possible wavelets Scales and positions, that is, an infinite
number of scales and positions. even though The discrete
wavelet transform (DWT) uses a finite set of wavelets, defined
as at a specific set of scales and locations The wavelet
transform is also a Mathematical tools for signal analysis

In optics, high-pass and low-pass can have different

meanings depending on whether they refer to the frequency or Fc= 1/2*Pi*C
wavelength of light, since these variables are inversely
proportional. High-pass filters would act as low-pass
wavelength filters and vice versa. For this reason, it is a good
practice to label wave filters as low and high pass to avoid
confusion corresponding to the high and low pass frequencies.
Low-pass filters exist in many different forms, including
electronic circuits such as a noise filter used in audio, anti-
aliasing filters for conditioning signals before analog-to-digital
conversion, digital filters for smoothing data files, acoustic
barriers, image blurring, and so on. The moving average
operation used in fields such as finance is a special kind of low-
pass filter and can be analyzed using the same signal processing
techniques used for other low-pass filters. Low-pass filters
provide a smoother signal shape, removing short-term
fluctuations and leaving a longer-term trend.

IJISRT22DEC763 www.ijisrt.com 190

Volume 7, Issue 12, December – 2022 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
ii) Testing a linear SVM:
• The separator is defined as the set of points for which:

Fig. 2 Wavelet Transform

C. Support Vector Machine

Support Vector Machine or SVM is one of the most
popular supervised learning algorithms used for both
classification and regression problems. However, it is
primarily used for classification problems in machine learning.
The goal of the SVM algorithm is to create the best line or
Fig. 3 Support Vector Machine
decision boundary that can segregate the one-dimensional
space into classes so that we can easily assign a new data point
VI. CONCLUSION
to the correct category in the future. This best decision
boundary is called the hyperplane. SVM selects extreme
The integration properties of all used hardware
points/vectors that help in creating the hyperplane. These
components were developed in it. The presence of each
extreme cases are called support vectors, which is why the
module has been carefully reasoned and placed, which
algorithm is called a support vector machine.
contributes to the best functioning of the unit. Our project
"Detection of physical exertion". It is primarily intended to
SVM can be of two types:
design a system that provides very accurate results than
1) Linear SVM: Linear SVM is used for linearly separable
existing equipment in today's world.
data, which means if a data set can be classified into two
classes using a single straight line, then such data is referred to
The microcontroller is programmed to receive input from
as linearly separable data and the classifier is used as Linear
the heart rate sensor when a finger is inserted and continuously
SVM. classifier.
display the value on the waveform. Today, many industries use
2) Non-linear SVM: Non-linear SVM is used for non-linearly
embedded systems for process control. In industries, we
separated data, which means if a data set cannot be classified
design embedded systems to perform specific operations such
using a straight line, then such data is referred to as non-linear
as monitoring temperature, pressure, humidity, voltage,
data and the classifier used is called a non-linear linear SVM
current, etc. and control other equipment based on these
classifier.
monitored levels.
i)Training a Linear SVM:
ACKNOWLEDGEMENT
• To find the maximum margin separator, we have to solve the
following optimization problem:
We express our profound gratitude to the principal of our
college Prof. S.D. Lokhande and to the Head of the
Department Information Technology Dr. S.R. Ganorkar for
allowing us to proceed with project and for giving us full
freedom to access the lab facilities. Our heart full thanks to our
guide Prof. Tejal. H. Patil for taking time and helping us
It's complicated, but it's a convex problem. There is only throughout our project. And, thanks to the staff member of IT
one optimum, and we can find it without worrying about Department. And last but not the least, a hearty thanks to all
learning speed, losing weight, or stopping early. – Don't worry our classmates for supporting me throughout the project.
about optimization problems.

It was resolved. This is called quadratic programming. –

Takes time proportional to N^2, which is bad for very large
datasets. So,for large datasets, we end up doing approximate
optimization!

IJISRT22DEC763 www.ijisrt.com 191

Volume 7, Issue 12, December – 2022 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
REFERENCES

[1]. S. Vishnu, S. R. J. Ramson, R. Jegan (2020). Internet of

medical things (IoMT). International Conference on
Devices Circuits and Systems (ICDCS) 101-104. Doi:
https://doi.org/10.1109/ICDCS48716.2020.243558
[2]. P. S. Akram, S. A. S. Valiveti, M. Ramesha, S. Sohail, K.
T. S. S. Rao (2021). IoT based remote patient health
monitoring system. 7th International Converence on
Advanced Computing & Communication System
(ICACCS) 978-1-6654-0521-8/20/$31.00.Doi:
https://doi.org/10.1109/ICACCS51430.2021.9441874
[3]. D. Gupta, A. Parikh, R. Swarnalatha (2020). Integrated
healthcare monitoring device for obese adults using
internet of things (IoT). International Journal of Electrical
and Computer Engineering (IJECE) 10 (1) 1239-1247.
Doi: https://doi.org/10.11591/ijece.v10i1.pp1239-1247
[4]. I. D. G. H. Wisana, B. Utomo, F. Amrinsani, E. Purwanto
(2021). Low coast health monitoring system based on
internet of things using email notification. International
Journal of Electronics, Electromedical, and Medical
Informatics (IJEEEMI) 3 (2) 45-52. Doi:
https://doi.org/10. 35882/ijeeemi.v3i2.2
[5]. K. T. Kadhim, A. M. Alsahlany, S. M Wadi, H. T.
Kadhum (2020). Monitoring vital signs of human hear
based on IOT 1 (2) 9-13. Doi:
https://doi.org/10.46649/110420.
[6]. Alamsyah, M. Subito, M. Ikhlayel, E. Setijadi (2020).
Internet of things based vital sign monitoring system.
International Journal of Electrical and Computer
Engineering (IJECE) 10 (6) 5891-5898.
[7]. S. Hadiyoso, A. Alfaruq, R. Tulloh, Y. S. Rohmah, E.
Susanto (2021). Internet of things based real-time vital
sign monitoring system using mobile application. Journal
of Applied Engineering Science 3 (19) 807-813.
[8]. P. Kalamkar, P. Patil, T. Bhongale, M. Kamble (2018).
Human Health Monitoring System using IoT and
Raspberry Pi. International Research Journal of
Engineering and Technology 5 (3) 3551-3553.
[9]. H. A. E. Zouka and M. M. Hosni (2019). Secure IoT
communications for smart healthcare monitoring system.
Journal Internet of Things 2542-6605 1-14. Doi:
https://doi.org/10.1016/j.iot.2019.01.003
[10]. G. Sheela and A. R. Varghese (2020). Machine Learning
based Health Monitoring System. Proceedings Materials
24 1788-1794.
[11]. S. Ahammed, N. Hassan, S. H. Cheragee, A. Z. M. T.
Islam (2021). An iot-based real-time remote health
monitoring system. International Journal of Research
Engineering Science 8 (3) 23-29. Doi:
https://doi.org/10.14445/23497157/IJRES-V8I3P104.

IJISRT22DEC763 www.ijisrt.com 192