VISVESVARAYA TECHNOLOGICAL UNIVERSITY

Jnana Sangama, Belgaum- 590 018

SYNOPSIS ON

“STRESS DETECTION USING MACHINE LEARNING”

ARTIFICIAL INTELLIGENCE AND DATA SCIENCE

During the Academic year 2023-2024
Submitted By
Student Name :P. VARSHAA
USN:1DB21AD033

Under the Guidance of

Mrs.MEGHANA H M
Assisstent Professsor
Dept. of AI&DS

Department of Artificial Intelligence and Data Science

Don Bosco Institute of Technology

Mysore Road, Kumbalgodu, Bengaluru-560074
 Synopsis & Presentation Contents

• Title
• Introduction
• Problem Statement & Motivation
• Literature Survey
• Aim & Objectives
• Proposed system
• Methodology
• Requirement Specification
• Implementation Details
• Innovativeness and Usefulness
• References
 Synopsis
Project Title: “Stress Detection using ML”

Sl.no USN Name E-mail id Contact

number
1. 1DB21AD033 P. VARSHAA [email protected] 6363757366

Sl.no Guide Name Designation

1. Mrs. Meghana H M Assisstent
Professor
2. Dr. Gowramma G S HOD
Introduction:
Stress is defined as a state of psychological and physiological imbalance resulting from the
disparity between situational demand and the individual’s ability and motivation to meet those
needs. Stress can cause a person to face social, physical organizational and emotional problems.
We can identify the factors causing stress also detect if a person has stress using ML and IoT
technologies.

• Machine Learning(ML): Machine Learning is the field of Artificial Intelligence(AI)

that uses statistical techniques to give computer systems the ability to learn from data
without being explicitly programmed. Supervised learning problems in machine
learning can be further grouped into Regression and Classification problems. A
classification model attempts to draw some conclusion from observed values. Given
one or more inputs a classification model will try to predict the value of one or more
outcomes.

• Internet of Things(IoT): The Internet of things (IoT) is the network of physical devices,
vehicles, home appliances, and other items embedded with electronics, software,
sensors, actuators, and connectivity which enables these things to connect, collect and
exchange data. IoT devices can be used to enable remote health monitoring and
emergency notification systems. These health monitoring devices can range from blood
pressure and heart rate monitors to advanced devices capable of monitoring specialized
implants.
Problem Statement & Motivation:
• Stress management systems are essential to identify and address stress levels that can
disrupt our socioeconomic functioning. According to the World Health Organization
(WHO), one in four people experience stress, which can result in mental and
socioeconomic issues, poor work relationships, and depression, and in severe cases,
suicide. Counselling is crucial to help people cope with stress, and while stress cannot
be avoided, preventive measures can be taken to mitigate its effects. Currently, only
medical and physiological experts can determine whether someone is experiencing
stress. Traditional stress detection methods rely on self-reported answers, which can
be unreliable. Automated stress detection can minimize health risks and improve
societal welfare. Therefore, there is a need for a scientific tool that can automate
stress detection using physiological signals. Stress detection is an important social
contribution that potential to improve quality of life. As IT industries bring new
technologies and products to the market, stress levels in employees are also
increasing. While some organizations offer mental health services to their employees,
more needs to be done to address this issue.
Literature Survey:

1. S. K. Kanaparthi, S. P, L. P. Bellamkonda, B. Kadiam and B. Mungara,

"Detection of Stress in IT Employees using Machine Learning Technique," 2022
International Conference on Applied Artificial Intelligence and Computing
(ICAAIC), Salem, India, 2022, pp. 486-493, doi:
10.1109/ICAAIC53929.2022.9793195.
keywords: {Employee welfare;Visualization;Sensitivity;Machine learning
algorithms;Human factors;Machine learning;Predictive models;Stress
prediction;KNN classification;facial expressions}.

2. A. Bannore, T. Gore, A. Raut and K. Talele, "Mental stress detection using

machine learning algorithm," 2021 International Conference on Electrical,
Computer, Communications and Mechatronics Engineering (ICECCME),
Mauritius, Mauritius, 2021, pp. 1-4, doi:
10.1109/ICECCME52200.2021.9590847. keywords: {Mechatronics;Machine
learning algorithms;Military computing;Databases;Face
recognition;Psychology;Feature extraction;CNN;Facial Expression
Recognition;LBP}.

3. P. Bobade and M. Vani, "Stress Detection with Machine Learning and Deep
Learning using Multimodal Physiological Data," 2020 Second International
Conference on Inventive Research in Computing Applications (ICIRCA),
Coimbatore, India, 2020, pp. 51-57, doi: 10.1109/ICIRCA48905.2020.9183244.
keywords: {Stress;Machine learning;Biomedical
monitoring;Sensors;Electrocardiography;Physiology;Support vector
machines;photoplethysmography;stressors;accelerometer;dichotomous;sudo
motor nerve activity;convex optimization}.
Objective/Aim of the Project:
Detection of Stress using the Internet of Things for data collection by different modules and
Machine Learning techniques such as Supervised learning and image processing techniques.

Proposed system:
Block Diagram:
Methodology:
Data Flow Diagram:

Requirement Specification:
Basic requirement for this project is
• Wi-Fi module(IoT component),
• Pulse sensors(IoT component),
• Camera module(IoT component),
• Computer system with Python containing some of the required packages installed.
Implementation Details:
Stress detection is one of the important topics in machine learning. Due to the stress the
people will not happy and there will not doing their work correctly. Based on the following
values we can easily predict the values.

In a proposed system figure 1 shows that we have used the statistical dataset where we have
used the six attributes are Electrocardiogram, Electromyogram, Galvanic Skin Response Hand
and Foot, Respiration and Heart Rate. Based on these algorithms we will pre-process the data
and that data is divided into training dataset and testing dataset. The dataset is compare with
the threshold values and that value will compare and predict the person is in stress or not
stress. We have to collect the data from datasets and that data should be pre-processed.
Some of the pre-processed steps are we have to import the dataset, import the libraries, we
have to clean. Based on that data we have encode the data and that data is dividing into
training data should be 80% and testing data should be 20%.

A. Electrocardiogram: Electrocardiogram measures the electrical activity of the heart rate and
the rhythm. Electrocardiogram measures the normal resting rhythm and stressed rhythm.
B. Electromyogram: Electromyogram measures the electrical activity of the muscle. When
the person is in stress the muscle tension will also be increased.
C. GSR Hand and Foot Galvanic Skin Response measures the electrical activity of the sweat
secretion. When the person is in suffering from stress the sweat glands will also be increased.
The sweat glands are highest in hand and foot region of the human being.
D. Respiration When the stress is increase on that time the respiration is also varies.
Respiration is measured by breaths per minute.
E. Heart Rate Heart Rate is measure as beats per minute. The normal heart rate is 60-100
beats per minute. As showed in a figure 2 to prevent the stress detection using machine
learning approach.

• Step 1: Data collection means collecting the data in my project I have use the statistical
datasets where I have considered the six attribute based on these attributes we will collect
the data.

• Step 2: Data pre-processing where we will be cleaning the data and also see the any missing
data is available in the values are to be checked.

• Step 3: Feature extraction is done by the values where in the values we have encoded the
values means the original data is encoded into unreadable format. Because so many people
are hacking the information due to the hacking will lost all the information.

• Step 4: We have extract the features by some of the algorithms are decision tree having
root nodes and sub node, when we consider in my project the root node is the dataset further
it is divided into sub-nodes as attributes, further the attributes divided into trained data and
compare with threshold value that value predicts the person is in stress or not stress , Naïve
Bayes is used for probability classification we calculating the problems based on assumptions
and probability and K-Nearest Neighbour is used for classification technique and predicts the
nearest neighbour in the data.

• Step 5: The dataset is divided into training dataset and testing dataset, always the training
dataset is more than the testing datasets.

• Step 6: Based on the trained value we will test the result as 0 as not stressed and 1 is
stressed. When we consider the trained value we have to compare with the threshold value
based on these values we can predict and detect the person is in stress or not stress.
F. Confusion Matrix: In a confusion matrix the predicted values are measured by Precision and
accuracy.
G. Algorithms
1) Decision Tree Decision tree [Figure 3] is a representation of all the possible features to a
tree. Decision tree is divided into root node, Sub node, Splitting node and the terminal node
or child node. Decision tree is represented as tree like structure.
Each of the nodes are further nodes are divided into additional nodes.
A decision tree as further constituents is:
1. Root node: Root node represents the main node and it divides into so many sub nodes.
2. Splitting: The root node is split into one or more nodes.
3. Decision node: When sub node is further divided into sub node is called as decision Node.
4. Leaf node or Terminal node: Nodes do not split into terminal node.
5. Branch node or sub node: The sub node of a complete tree is called as sub node.
6. Parent and Child Node: Parent node is a root node, root node is further splits into sub node
and that node is divided into sub node are called as leaf or child node.
2) Naïve Bayes Classifier Naïve Bayes is one of the classification algorithms. Naïve
Bayes is one of the supervised learning techniques. Naïve Bayes is classified as a probability
classifier and it is based on the bayes theorem. Naïve Bayes is the simplest algorithms for
classification based on Bayes theorem. Naïve Bayes are used for large sets of data.
For example: When considered the apple as a fruit and also relate the entire feature because
it apple is in red colour, it is round shape, and also defined its diameter. So assume based on
the probability of getting the fruit is an apple or not apple.

3) K-Nearest Neighbour :
K-Nearest Neighbor is one of the important classification algorithms. K-Nearest Neighbor
stores all the feasible cases and classifies new cases based on a similar measure of the data.
KNearest Neighbor measures the nearest neighbour by the surrounding. K-Nearest
Neighbour is divided into classification and regression. The neighbours are taken from a
group of objects.
Working of K-Nearest Neighbour Algorithm:
K-Nearest Neighbours algorithm uses feature similarity to predict the values of new datasets
which is having new data point will be assigned a value based on the points in the training set.
• Step 1: For implementing the algorithm need to collect the dataset during the first step of
K-Nearest Neighbour and load the training as well as test data.

• Step 2: Choose the value of K i.e. the nearest data points and k can be any integer.

• Step 3: Calculate the distance between test data and each row of training data with the
help of method is to calculate the distance.

• Step 4: Based on the distance values, sort them in ascending order.

• Step 5: Choose the k rows for from the sorted array

. • Step 6: Assign a class to the test point is based on the class of each row.

• Step 7: End.

Innovativeness and Usefulness:

This project works for all kind of people with any facial expressions at any time with accuracy.
It is able to identify stress level of a person by using his facial expression and pulse rate. This
project will help a person to identify his stress so that he can take measures to reduce his stress.

References:
1. https://github.com/nandahkrishna/StressAffectDetection
2. L. Rachakonda, S. P. Mohanty, E. Kougianos, and P. Sundaravadivel, “Stress-Lysis:
A DNN-Integrated Edge Device for Stress Level Detection in the IoMT,” IEEE Trans.
Conum. Electron., vol. 65, no. 4, pp. 474–483, 2019.
3. L. Rachakonda, P. Sundaravadivel, S. P. Mohanty, E. Kougianos, and M.
Ganapathiraju, “A Smart Sensor in the IoMT for Stress Level Detection”, in
Proceedings of the 4th IEEE International Symposium on Smart Electronic
Systems (iSES), 2018, pp. 141--145.
4. Azzedine Idir Aitsaid, Mohamed BENRABAH, Wadjih Bencheikh. (2023).
Stress Detection. Kaggle.
https://kaggle.com/competitions/soaicommunitydatathon23-stress-detection.
5. https://www.youtube.com/watch?v=NUZ1_LKUM7E
6. https://www.youtube.com/watch?v=mItWe17RqxE
7. Z. Zainudin et al 2021 J. Phys.: Conf. Ser. 1997 012019