Final Documentation (Major Project)

A PROJECT REPORT
On
PARKINSON’S DETECTION SYSTEM
Submitted in the Partial Fulfillment of the Requirement for the Award of

BACHELOR OF TECHNOLOGY
In
Computer Science & Engineering (2024)
By
Astha Dixit [2000460100018]

Astha Mishra [2000460100019]
Janhavi Srivastava [2000460100032]
Priti Yadav [2100460108002]
Under the Guidance Of

Mr. Milind Bhatt
MAHARANA PRATAP ENGINEERING COLLEGE
Affiliated to
Dr. APJ ABDUL KALAM TECHNICAL UNIVERSITY,
UTTAR PRADESH, LUCKNOW
CERTIFICATE
Certified that the project entitled “PARKINSON’S DETECTION SYSTEM

”submitted by Astha Dixit [2000460100018], Astha Mishra [2000460100019] , Janhavi
Srivastava [2000460100032] ,Priti Yadav [2100460108002] in the partial fulfillment of
the requirements for the award of the degree of Bachelor of Technology (Computer
Science & Engineering) of Dr. APJ Abdul Kalam Technical University (Uttar Pradesh,
Lucknow), is a record of students’ own work carried under our supervision and guidance.
The project report embodies results of original work and studies carried out by students
and the contents do not form the basis for the award of any other degree to the candidate
or to anybody else.
Asst. Prof. (Mr.) Milind Bhatt Prof. (Dr.) Atul Kumar
(Project Guide) (Head of Dep
iiiiiiiii
COMPUTER SCIENCE & ENGINEERING
DECLARATION
We hereby declare that the project entitled “PARKINSON’S DETECTION SYSTEM”

submitted by us in the partial fulfillment of the requirements for the award of the degree
of Bachelor of Technology (Computer Science & Engineering)of Dr. APJ Abdul Kalam
Technical University (Uttar Pradesh, Lucknow), is a record of my/our own work carried
under the supervision and guidance of Mr. Milind Bhatt.
To the best of our knowledge this project has not been submitted to Dr. APJ Abdul
Kalam Technical University (Uttar Pradesh, Lucknow) or any other University or
Institute for the award of any degree.
Astha Dixit Astha Mishra

2000460100018 2000460100019
Janhavi Srivastava Priti Yadav

2000460100032 2100460108002
iiiiiiiii
ACKNOWLEDGEMENT
We would like to take this opportunity to express our gratitude towards all those people
who have in various ways, helped in the successful completion of our final year project
work on “PARKINSON’S DETECTION SYSTEM” done satisfactorily at
MAHARANA PRATAP ENGINEERING COLLEGE. We must convey our gratitude to
Mr. Milind Bhatt for giving us the constant source of inspiration and help in preparing the
project, personally correcting our work and providing encouragement throughout the
project. In that regard we are especially thankful to our Head of the Department Dr. Atul
Kumar for steering me through the tough as well as easy phases of the project in a result-
oriented manner with concerned attention. A special thanks to all the faculty members of
our CSE department.
We would also love to take this opportunity to tell the readers of our material that their
comments, be it appreciation or criticism would be the most valuable thing to us. That
will be something we will always be thankful for.
Date:
iiiiiiiii
PREFACE
Parkinson's disease is a complex neurodegenerative disorder that affects millions of people

worldwide. Early and accurate diagnosis of Parkinson's disease is crucial for effective treatment
and management of the condition. In recent years, there has been growing interest in using
machine learning techniques to aid in the diagnosis of Parkinson's disease, particularly using
medical imaging data.
This project focuses on the development of a Parkinson's disease detection system using spiral
images and machine learning algorithms, implemented in Python. Spiral drawing tests are
commonly used in clinical settings to assess motor function in individuals with Parkinson's
disease. These spiral images provide valuable insights into the progression of the disease and can
be used as a diagnostic tool.
The main objective of this project is to develop a machine learning model that can accurately
discriminate between healthy control subjects and those with Parkinson's disease based on
features extracted from spiral images. We aim to leverage the power of convolutional neural
networks (CNNs) to extract meaningful features from the spiral images, which will then be used
to train a classification model.
To enhance the performance of our model, we will utilize the SCI-KIT Image Data Generator API
for data augmentation during training. Data augmentation techniques such as rotation, flipping,
and zooming will be applied to increase the diversity of our dataset, thereby improving the
generalization capabilities of our model.
iiiiiiiii
This project represents a significant step forward in the field of Parkinson's disease diagnosis, as it
demonstrates the potential of machine learning and image analysis techniques to assist clinicians in
early and accurate detection of the disease. We hope that our work will contribute to the
development of more effective diagnostic tools for Parkinson's disease, ultimately leading to better
outcomes for patients.
iiiiiiiii
ABSTRACT
Parkinson’s disease (PD) is one of the major public health diseases in the world which is
progressively increasing day by day and has had its effect on many countries. Thus, it is very
important to predict it in early age which has been a challenging task among researchers because
the symptoms of disease come into existence in either middle or late middle age. So this Project
focuses on the Spiral Test difficulty symptoms of PD affected people and formulates the model
using various machine learning techniques such as adaptive boosting, RNN, Convolutional Deep
Neural Networks, support vector machine, decision tree, Convolutional Neural Networks and
linear regression. Performance of these classifiers is evaluated using various metrics i.e.
accuracy, receiver operating characteristic curve (ROC), Sensitivity, precision, specificity. At
last, Feature selection technique is used to find the most important features among all the feature
to predict the Parkinson’s disease
iiiiiiiii
TABLE OF CONTENTS
Certificate ii
Declaration iii
Acknowledgement iv
Preface v-viii
Abstract ix-x
Table of Contents xi-xii

1. Introduction 1-4
2. Objectives 5-6
3. Work Proposal 7-15
4. Literature Review 16-23

5. Proposed Methodology 24-30
5.1 Data Flow Diagram (Level 0)
5.2 Data Flow Diagram (Level 1)
5.3 ER Diagram
5.4 Flow Chart
5.5 Structure Chart
6. Hardware and Software Requirements 31-32
6.1 Hardware Requirements
6.2 Software Requirements

7. Database Requirements 33-34
8. Technology Used 35-36
iiiiiiiii
8.1 Frontend Technologies
8.2 Backend Technologies
9. Implementation 37-41
10. Result 41-52
11. Conclusion and future scope of the project 53
 Appendices
 Appendix A x-xi
xii
 Appendix B
xiii -xxii
 Appendix C xxiii - xxv
 Appendix D xxvi- xxviii
 References
 Research paper
iiiiiiiii
Parkinson Detection System
CHAPTER 1 : INTRODUCTION
Neurodegenerative disorders are the results of the progressive tearing and neuron loss in
different areas of the nervous system. Neurons are the functional unit of the brain .They are
contiguous rather than continuous. A good healthy looking neuron as shown in fig 1 has
extensions called dendrites or axons, a cell body and a nucleus that contains our DNA. DNA is
our genome and hundred billion neurons contains our entire genome which is packaged into it
.When a neuron get sick, it loses its extension and hence its ability to communicate which is not
good for it and its metabolism become low so it starts to accumulate junk and it tries to contain
the junk in the little packages in little pockets .When things become worse and if the neuron is a
cell culture it completely loses its extension, becomes round and full of the vacuoles.
Fig 1: Structure of neuron present in human brain
This work deals with the prediction of Parkinson’s disorder which nowadays is a tremendously
increasing incurable disease. Parkinson’s disease is most spreading disease [19] which get its
CSE DEPARTMENT,MPEC,Kanpur 1
name from James Parkinson who earlier described it as a paralysis agitans and later gave his
surname was known as a PD. It generally affects the neurons which are responsible for overall
body movements. Main chemicals are dopamine and acetylcholine which affect the human brain.
There are various environmental factor which have been implicated in PD [20].below are the
listed factor which caused Parkinson’s disease in an individual.
A. Environmental factors: Environment is defined as the surrounding or the place in which an

individual lives .So environment is the major factor that will not only affect the human’s
brain but also affects all the living organisms who live in the vicinity of it. Research and
evidence has proved that the environment has a big hand in the development of
neurodegenerative disorders mainly Alzheimer’s and Parkinson’s.
There are certain environmental factors that are influencing neurodegenerative disorder
with high pace are:-
a. Exposure to heavy metals (like lead and aluminum) and pesticides.

b. Air Quality: Pollution results in respiratory diseases.
c. Water quality: Biotic and Abiotic contaminants present in water leads to water pollution.
d. Unhealthy lifestyle: It leads to obesity and a sedentary lifestyle.
e. Psychological stress: It increases the level of stress hormone that depletes the functions
of neurons.
B. Brain injuries or Biochemical Factors: Brain is a control center of our complete body. Due
to certain trauma people have brain injuries which leads some biochemical enzymes to come
into picture which provides neurons a stability and provide support to some chromosomes
and genes in maintenance.
C. Aging Factor: Aging is one of the reasons for the development of Parkinson's disease. Wi.
According to the author [22], in India, 11,747,102 people out of 1, 065, 070, 6072 are
affected from Parkinson’s disease.
D. Genetic factors: Genetic factor is considered the main molecular physiological cause which
leads to neurodegenerative disorders. The size, depth and effect of actions of different genes
defines the status or level of neurodegenerative disease which increases itself gradually over
time. Mainly the genetic factors which lead to Neurodegenerative disorders are categorized
into pharmacodynamics and pharmacokinetics [21].
E. Spiral Test factors: Due to the condition associated with Parkinson’s disease (rigidity and
bradykinesia), some speech language pathology such as Drawing , articulation and
swallowing alterations are found . There are various ways in which Parkinson’s disease (PD)
might affect the individual.
(i) The Drawing gets breathy and softer.
(ii) Speech may be smeared.
(iii) The person finds difficulty in finding the right words due to which speech becomes slower
➢ Motivation
Ten percent of people aged 65 or more do have a neurodegenerative disease, and there are no
cures for them. Almost 30% of the people are facing this incurable disease[23]. Current
treatment, if available at all,only reduces symptoms and that too for a limited period of time. The
main cause for Parkinson's disease is the accumulation of protein molecules in the neuron which
gets misfolded and hence causing Parkinson’s disease. So till now researchers got the symptoms
and the root cause i.e. from where this disease had evolved. But very few have come to its
cure.So in this era where Parkinson's disease is progressing at double pace ,it is very important to
find the solution which can detect it in its early phases.
Parkinson’s Disease Symptoms
The symptoms of Parkinson disease are broadly divided into two categories.
● Motor symptoms
● Non-motor symptoms
Motor symptoms
This is a symptom where any voluntary action is involved. It‘s indicates the movement
related disorders like tremor, rigidity, freezing, Bradykinesia or any voluntary muscle
movement
Non-motor symptoms
Non motor symptoms include disorders of mood and affect with apathy, cognitive dysfunction as
well as complex behavioral disorders. There are two other categories of PD which are divided by
doctors: Primary symptom and Secondary symptom.
Primary symptoms
It is the most important symptom. Primary symptoms are rigidity, tremor and slowness of
movement.
Secondary symptoms
It is a symptom which directly impacts the life of an individual. These can be either motor or
non- motor. Its effect depends on person to person.. A very wide range of symptoms is associated
with Parkinson's,.diseases
Besides these symptoms, there are some other symptoms found that lead to Parkinson’s disease.
These symptoms are micrographia, decreased olfaction & postural instability, slowing of the
digestive system, constipation, fatigue, weakness and Hypotension [24]. Speech difficulties i.e.
dysphonia (impaired speech production) and dysarthria (Spiral Test difficulties) [25] are found in
patients of parkinsons..
CHAPTER-2 : OBJECTIVES
The project objective of a Parkinson's detection system is to develop a technology or system

that can assist in the early detection, monitoring, or management of Parkinson's disease.
Parkinson's disease is a neurodegenerative disorder that affects movement and can have a
significant impact on a person's quality of life. Early detection and intervention can lead to
better treatment outcomes and improved quality of life for individuals with Parkinson's.
Here are some specific objectives that a Parkinson's detection system might aim to achieve:
Early Diagnosis: Develop a system that can detect early signs and symptoms of Parkinson's
disease, potentially even before noticeable motor symptoms occur. This could involve the use
of advanced technologies such as machine learning algorithms, wearable devices, or other
diagnostic tools.
Objective Monitoring: Create a system that can objectively monitor the progression of
Parkinson's disease over time. This could include tracking changes in movement patterns,
tremors, speech, or other relevant metrics.
Remote Monitoring: Enable individuals with Parkinson's to be monitored remotely, reducing

the need for frequent in-person visits to healthcare facilities. This could improve the
convenience and accessibility of care, particularly for those who may have difficulty traveling.
Personalized Treatment Plans: Use collected data to tailor treatment plans for individuals
with Parkinson's. This could involve optimizing medication dosages, suggesting specific
exercises or therapies, or providing lifestyle recommendations.
Predictive Analytics: Develop algorithms that can predict the future progression of the disease
based on current data. This could help healthcare providers and patients make informed
decisions about treatment and lifestyle adjustments.
Improved Quality of Life: Ultimately, the goal of a Parkinson's detection system is to improve
the quality of life for individuals living with the disease. This could be achieved by providing
timely interventions, reducing symptom severity, and minimizing the impact of the disease on
daily activities.
Research and Data Collection: Contribute to the broader understanding of Parkinson's disease
by collecting anonymized data from a large number of individuals. This data can be used for
research purposes, potentially leading to new insights into the disease and its treatment.
User-Friendly Interface: Ensure that the system is user-friendly and accessible to both
healthcare professionals and patients. This may involve designing intuitive interfaces providing
clear instructions, and offering support resources.
Compliance and Safety: Ensure that the system complies with relevant medical and privacy
regulations to protect patient data and ensure the safety and effectiveness of the technology.
Cost-Effectiveness: Consider the cost implications of the system, aiming to provide an

affordable solution that can be widely accessible to those who need it.
It's important to note that any project related to healthcare, especially one involving diagnosis or
treatment, should be developed and implemented with the guidance and oversight of medical
professionals and in accordance with regulatory standards and ethical considerations
CHAPTER 3 : WORK PROPOSAL
This chapter deals with the description of the dataset used and the approaches taken to achieve
the early prediction of Parkinson’s disease in a PD patient .The approaches taken were selected
with the aim to distinguish a Parkinson’s disease patient from those who are healthy patient. The
idea is to do a comparative analysis of different machine learning techniques by implementing
different models on the selected dataset and finding the best machine learning technique among
them by evaluating some performance metrics like accuracy, ROC, AAE, and ARE etc. Further
the work is extended by implementing Feature selection technique.
ARCHITECTURE
Dataset Description
The PD and control handwriting database consists of 62 PWP (People with parkinson) and 15
healthy individuals who appealed at the Department of Neurology in Cerrahpasa Faculty of
Medicine, Istanbul University. From all subjects, three types of handwriting recordings (Static
Spiral Test (SST), Dynamic Spiral Test (DST) and Stability Test on Certain Point (STCP)) are
taken. Also the drawings of spirals belonging to the PWP are included in the dataset as image.
Therefore, this dataset can also be used for regression.
We used a public dataset: the Parkinson Disease Spiral Drawings Using Digitized Graphics Tablet
dataset [27]. This dataset includes spiral drawings from 77 people, 62 with PD and 15 healthy
angle (Figurepeople in the control group. The dataset was recorded using the Wacom Cintiq 12WX
graphics tablet [29]. This tablet allows showing a PC’s screen on its display and interacting with a
digital pen. In total, five time-series were recorded for each drawing. Every recording includes
information about X-Y-Z coordinates, pressure, and grip 1) obtained from the API functions of
the device.
Handwriting dataset was constructed using the Wacom Cintiq 12WX graphics (Hahne et al., 2009)
table. It is basically a graphics tablet and LCD monitor rolled into one. It is able to display a PC's
screen on its monitor and only interacts with digitized pens. Special software was designed for
recording handwriting drawings and testing the coordination of the PD patients using the
recordings. The software uses API functions of the device and was developed in C# platform which
can be run on Windows systems You can contact the authors to request the software which is
mentioned [1]. In this study, there are three different kinds of tests developed for the data collection
via graphics tablet. The first one is the Static Spiral Test (SST) which is frequently used for clinical
research in the literature for different purposes like determining motor performance (Wang et al.,
2008), measuring tremor (Pullman, 1998) and diagnosing PD (Saunders et al., 2008). In this test,
three wound Archimedean spirals appear on the graphics tablet using the software and patients
were asked to retrace the same spiral as much as they can using the digital pen. During the test, the
features which are mentioned above and the other data to specify the patient are recorded to the
database. The second test is the Dynamic Spiral Test (DST). Unlike SST, the Archimedean spiral
just appearsand disappears in certain time intervals, in other words the Archimedean spiral blinks.
This forces the patient to keep the pattern in mind and continue to draw. The purpose of this test is
to determinethe change in patient's drawing performance and pause times since it is more difficult
to retrace the Archimedean spiral in this case. As a result of this test, it is observed that most of the
patients continued drawing but nearly all of them lost the pattern.
The third test is the Stability Test on Certain Point (STCP). In this test, there is a certain red point
in the middle of the screen and the subjects are asked to hold the digital pen on the point without
touching the screen in a certain time. The purpose of this test is to determine the patient's hand
stability or hand tremor level
Fig 2: Sample dataset of biomedical Drawing measurements of 31 people
ALGORITHMS
Deep Neural Networks
Deep Neural Networks had its base as that of biological neurons which is used for prediction..
Let’s understand the single neuron. In fig 3 you can see a diagram of a single neuron with single
input. The given equation will explain the single input neuron where O is the output ,σ is the
sigmoid function or transformed function ,ξ is the input to the neuron and ω is the weight that
connects that input to the neuron
O = σ(ξ ω)...................................................................................................................................|
Fig 3: A Single input neuron
So when multiple inputs are given to a neuron as mentioned in fig 4, it will form a MLP.which
consists of inputs connected through the weights in the form of layers. So the neuron takes
multiple inputs and generates output which is known as Multilayer perceptron. The diagram
below demonstrates a multilayer perceptron.
Fig 4 : Multilayer perceptron
O = 𝛔(𝛏𝟏 𝛚𝟏 + 𝛏𝟐 𝛚𝟐 + ⋯ … … + 𝛏𝐤𝛚𝐤) +Θ...........................................................................2

where O is the output
o is the sigmoid function or transformed function
ξ is the input to the neuron
ω is the weight of input (1 to k)

Θ is the bias
Linear Regression
This model is used to find relationships between two continuous variables. One variable is
called the dependent or response and the other one is called the independent or predictor
using a best fit straight line known as the regression line. The purpose of the linear
regression model is that it looks for a statistical relationship between the two variables and
not the deterministic variable. By deterministic relationship we mean that if one variable can
be accurately expressed by the other one.
Fig 5: Straight line plot in Linear regression

The mathematical representation of Linear Regression:
Y=[X][W]+ [B]....................................................................................................3
Y=b0 +b1X1 +b2X2..............................................................................................................................................4
In eq 3 and eq 4, Y is the dependent variable and X1,X2 represents the independent

variables b1, b2 are the coefficients of the independent variables and b0 is the intercept .
Convolutional Neural Networks
Convolutional Neural Networks is one of the machine learning methods which is used for both
classification as well as regression tasks. It is a type of ensemble method with which a group of
weak models when combined turns into a powerful model. In Convolutional Neural Networks,
multiple trees are created .To classify every tree gives a classification, and are supposed to vote
for that class. The forest selects the classification having the highest votes. The selection process
by Convolutional Neural Networks is shown in fig 6.
Convolutional Neural Networks Prediction Pseudo code:
1. Takes the test sets features and makes decision trees to predict the outcomes
and stores the predicted outcomes.
2. Calculate the votes for each predicted outcome.
3. Consider the high voted predicted outcome as the final prediction.
Test sets features
All Trees
Prediction
Tree 1
Tree 2
Random Forest
Fig 6: Prediction process taken by Convolutional Neural Networks
Decision Tree
Decision tree algorithm is a supervised learning algorithm which is used for classification as well
as regression problems. The main objective of using Decision tree is to create a training model
which can be used for prediction of Parkinson’s by learning decision rules inferred from training
datasets. It tries to resolve the problem by using tree representation or tree hierarchy.It has three
nodes:
1. Root
2. Internal Node
3. Leaf nodes
ROOT
Internal Internal
nodes nodes
Le Le Le
Fig 7 : Representation of decision tree
Root node represents the entire sample which is further splits into nodes known as leaf nodes
which represents the attribute which is further divided into leaf nodes which represents the class
labels.
Ad boost
Ad boost, like the Convolutional Neural Networks classifier, is another ensemble classifier.
AdaBoost which is known as adaptive boosting which is used for classification rather than
regression .It is a best algorithm for predicting. It is used to boost the performance of decision
tree or binary classification problems.
. For the new input we are providing to adaboost, each weak learner calculates a predicted value
.the value can be either 1.0 or -1.0. Each weak learner weights the predicted values. The
prediction for the ensemble model is calculated by taking the sum of the weighted predictions.If
the Sum is
positive it will be assigned the First predicted class,if Sum is negative it comes under Second
predicted class.
Mathematics involved in Adaboost :
H(x) = sign( ∑ ah(x)
t=1
h(x) is the output of weak classifier t for input

x a is the weight assigned to the classifier.
a = 0.5 * ln ((1-E)/E)
Support Vector Machine

Support vector machine is defined by separating hyperplanes. The output of the approach is an
optimal hyper-plane which categorizes new examples. In 2 dimensional space, this new hyper
plane is a line dividing a plane in two parts where each class lies on one side.It gives better
results for complex classification problems. Each data item is plotted as a point in n- dimensional
space with the value of every feature reflecting the coordinates of the plane. The SVM performs
classification that differentiating the two classes very efficiently.
Fig 8: Hyper plane classifying two classes.
RNN
RNN is an ensemble algorithm, RNN methods form an efficient class of algorithms which bring
together several instances of black box estimators on random subsets of the original data set and
then efficiently aggregate their individual predictions to process and formulate the final
prediction. The RNN methods make immense efforts to reduce the variance of the base estimators
byefficiently introducing the randomization into its construction and then making an ensemble
from it. Let’s take an instance where you have a learner for example The Decision Tree. Many
times you have made efforts to improve its accuracy and variance by applying Bootstrap
technique.
You end up generating multiple samples from your data set that has been classified as a training
set using an approach of the next scheme: you can take randomly any element from your training
set and then can pull it back. This results in a scenario where some of the elements of the training
set will be present multiple times in the generated new sample and some will accidentally be
absent. These samples should have the same size as the train set.
1. You can train your learner on each generated sample to gain the efficient results and improve
the model better.
2. When you apply the algorithm you are just doing an average prediction of learners in case of
regression or make the voting in case of classification.
CHAPTER 4 : LITERATURE REVIEW
Previous studies to predict PD have been implemented on MRI scans, gait and genetic data, but
research on audio impairment for early detection is minimal. For instance, Bilal et. al. [7]
studied genetic data to predict the onset of PD in senior patients with SVM model. They trained
an SVM model to reach an accuracy of 0.889, while this research paper describes an improved
SVM model with an accuracy of 0.9183. These results also corroborate the merits of
classification of PD based on audio data, over genetic data.
1. “Pathoanatomy of Parkinson’s disease” a book written by Braak, H., Braak, E. has

garnered significant attention and acclaim since its publication in 2000. The study of the
causes of disease based on the examination of organs and tissues which was concerned
about the Parkinson's diseases and gives the Introduction and early detection of
Parkinson's disease and gives its symptoms and causes along with its prevention and
remedies.This approach reveals a key role by the amygdala and related structures in extranigral
pathology. Severe lesions occur in the central amygdaloid nucleus, in nuclei projecting to the
cerebral cortex in a non-specific manner, and in nuclei regulating endocrine and autonomic
functions. It is suggested that extranigral lesions contribute to the development of behavioral
changes and autonomic dysfunction
2. Raundale, Thosar and Rane [8] used keystroke data from UCI telemonitoring dataset to
train a Random Forest classifier to predict the severity of PD in older patients.People have
trouble vocally, writing, strolling, or completing other simple tasks when dopamine-
generating neurons in parts of the brain become impaired or expire. These symptoms
worsen over time, increasing the severity of the condition in patients. We have suggested a
methodology in this article for the prediction of Parkinson's disease severity using deep
neural networks on UCI's Parkinson's Telemonitoring Vocal Data Set of patients. We have
created a neural network to predict the severity of the disease and a machine learning
model to detect the disorder. Classification of Parkinson's Disease is done by Neural
network, Random Forest Classifier.
3. Cordella et. al. [9] use audio data to classify PWP, however their models are heavily reliant
on MATLAB. Our research uses open-source models trained in Python, that are faster and
memory efficient. Majority of research done emphasizes the use of deep learning in
PDdetection.
4. Ali et. al. [10] who explain the use of ensemble deep learning models applied to phonation
data, to predict the progress of Parkinson’s disease. Their work lackedthe use of feature
selection that would improve Deep learning model (DNN) performance. Hence, this paper
implements PCA on 22 attributes to select 7 major voice modalities in PDdetection.
5. Huang et. al. [11] aim to reduce PD diagnosis dependence on wearable equipment by traininga
traditional decision tree on 12 complex speech features of the MDVR-KCL [12] dataset. The
devices that NICE has recommended are Kinesia 360, KinesiaU, PDMonitor, Personal
KinetiGraph (PKG), and STAT-ON. Read more about the guidance on the NICE website.
6. Wodzinski et. al. [13] trained a ResNet model on images of audio data, instead of training the
model on the nuances of the frequency of audio. ResNet-50 is a convolutional neural network
that is 50 layers deep. You can load a pre trained version of the neural network trained on
more than a million images from the ImageNet database. The ResNet model architecture
allows the training error to be reduced with a deeper network through connection skip.
ResNet stands for Residual Network and is a specific type of convolutional neural network
(CNN) introduced in the 2015 paper “Deep Residual Learning for Image Recognition” by He
Kaiming, Zhang Xiangyu, Ren Shaoqing, and Sun Jian. CNNs are commonly used to power
computer vision applications.
7. Wroge et. al [14] aimed to remove subjectivity of doctors in prediction of PD using an

unbiased ML model, however their results achieved peak accuracy of 85% only. The system
processes the symptoms provided by the user as input and gives the output as the probability
of the disease. Naïve Bayes classifier is used in the prediction of the disease which is a
supervised machine learning algorithm. The probability of the disease is calculated by the
Naïve Bayes algorithm is 86.7
8. Wang et. al. [15] implemented 12 machine learning models on 401 voice biomarkers
dataset to classify patients as PD or not. They built a custom deep learning model (DEEP)
with classification accuracy of 96.45%, however the model was expensive due to large
memory requirements.AI, especially ML and DL methods, could enhance the accuracy of
predicting genetic disorders. For instance, the ANN-based model achieved 85.7%, 84.9%,
and 84.3% for the training, testing, and validation phases, respectively . The ML
performance accuracy in psychiatry from genotypes varied between 48 and 95%. I can
improve patient recruitment by identifying and screening potential participants based on
inclusion and exclusion criteria. This can help reduce the time and cost associated with
patient recruitment, which is a common clinical research issue.
9. Alkhatib et. al. [16] implemented a linear classification model with 95% accuracy to
characterize shuffling movement of PD patients. Their study focused on gait of patients and
future work encouraged the use of audio and sleep data to improve the results. Majority of
research done emphasizes the use of deep learning in PD detection, such as, Aditi Govindu et.al
who explains the process of predicting the progress of PD by using e KNN, logistic regression,
random forest regression and SVM models and got 91.83% of success rate. Ali et. al. who
explains the use of ensemble deep learning models applied to phonation data, to predict the
progress of Parkinson’s disease. Their work lacked the use of feature selection that would
improve Deep learning model (DNN) performance. Hence, this paper implements PCA on 22
attributes to select 7 major voice modalities in PD detection. Huang et. al. aim to reduce PD
diagnosis dependence on wearable equipment by training a traditional decision tree on 12
complex speech features of the MDVR- KCL dataset. Wodzinski et. al. trained a ResNet model
on images of audio data, instead of training the model on the nuances of the frequency of audio.
Wroge et. al aimed to remove subjectivity of doctors in prediction of PD using an unbiased ML
mo
10. Ricciardi et. al [17] performed spatial-temporal analysis of brain MRI scans. They
implemented decision trees, random forest and KNN to detect Mild Cognitive Impairment (MCI)
in PWP. However, dataset was small and artificial data augmentation [18] was needed. Neural
network, support vector machine (SVM), k-nearest neighbor (KNN) algorithm, ensemble, and
regression models are the various ML techniques that are used to distinguish patients with AD
and mild cognitive impairment (MCI) and healthy controls (HCs). Previously, there have been
numerous efforts to evaluate the result using a monoclassifier, but due to some factors,
satisfactory accuracy was not obtained, so various experiments were performed to evaluate the
accuracy using a multiclassifier or ensemble method of the classifier.SVM is a multivariate,
supervised data classification approach. Its main purpose is to provide the optimal hyperplane
that divides data points of one class from another class.
The utilization of SVM techniques in neuroimaging scans has depicted the potential to predict
future cognitive deterioration and transformation from MCI to AD. It can conceivably be
employed in multimodal neuroimaging scans to improve the accuracy of AD diagnosis.
11. U. Haq and colleagues [19] implemented L1-support SVM, without feature identification on
vowel phonation dataset for neurological disorder patients. Their paper focused on patient age
group of 46-85 years, Aditi Govindu et al. / Procedia Computer Science 218 (2023) 249–261
251
Author name / Procedia Computer Science 00 (2019) 000–000 3 without considering healthy
individuals in a lower age bracket. Traditionally, in the absence of technological methods,
individuals with AD were cared for by their family members. Reviews show that 68% of
caregivers of patients with dementia have burden and stress in the procedure of caregiving and
65% develop symptoms of depression. Alternatives to augment care with the use of robots can
help reduce this stress. There are SARs: Robots that are trained to assist patients with AD by
the means of social interactions are used , and the rehabilitation process that they follow
involves the crucial factor of assessing the patient’s movement rate and intensity, associated
with their involvement with the robots.
12. Mei et. al. [20] explain the importance of ML to detect PD, as subtle non-motor symptoms can
be missed during subjective evaluation by a doctor. Their work reviews 209 studies based on
dataset, ML methods and outcomes achieved. Mitchell, T.M. Machine Learning; McGraw-
Hill: New York, NY, USA, 1997 The field of machine learning is concerned with the question
of how to construct computer programs that automatically improve with experience. In recent
years many successful machine learning applications have been developed, ranging from data-
mining programs that learn to detect fraudulent credit card transactions, to information-
filtering systems that learn users' reading preferences, to autonomous vehicles that learn to
drive on public highways. At the same time, there have been important advances in the theory
and algorithms that form the foundations of this field.
13. Biomedical voice measurement [24] of 31 people have been gathered, where 23 patients
have PD. Patients are in the age range of 46 to 85 years, while normal readings are from
people of 23 years of age. An average of 6 phonation’s were recorded 195 times for every
person, ranging from 1 to 36 seconds in duration.
14. Data wrangling [26] is implemented to clean data and handle missing attributes in the dataset.
depicts the noise to harmonic tone (NHR) ratio and harmonic tone to noise ratio (HNR) for
PWP. As stages of the disease progress, noise in speech increases resulting in increased NHR,
as seen
. The skew in the data and low value of NHR (0.3) indicates poor voice quality.
15. Aditi Govindu et al. / Procedia Computer Science 218 (2023) 249–261 6 Author name /
Procedia Computer Science 00 (2019) 000–000 depicts a box plot of all 22 attributes in the
dataset. It depicts the spread and skewness of data over a median quartile. Figure depicts blue
as normal records and orange as PWP records. NHR data points for PWP have maximum no.
of outliers, due to greater noise in speech. Similarly, HNR records have maximum data outliers
below the median, for PWP records.. It is used to highlight the shift in shimmer of voice for
PWP,compared to healthy patients. It shows that Shimmer: APQ3 and Shimmer: DDA are
linearly correlated while
16. Logistic regression [27] is a prevalent supervised ML algorithm that predicts categorically
dependent variables using a set of independent variables. It uses a curve fitting method to
predict a probabilistic value in the range of 0 to 1, as the outcome of a categorical or discrete
input. Compared to Linear regression [28], where a line is fit to linearly predic
values in range 0 to 1. This is ideal for audio data, as attributes affecting classification of PD are
not linearly correlated, rather follow an exponential pattern.
17. . Random Forest classifier - Random Forest is a supervised machine learning algorithm that is
applicable to classification and regression problems. This paper implements random forest
classifier [29] to train a number of decision trees on subsets of the dataset and consider the
average to increase the predictive accuracy of the results. It is a democratic model, where no
single decision tree model is treated as superior, instead the majority vote of prediction from
all models is considered to give an average prediction of the output. As no. of trees increases,
the chances of overfitting decrease.
18. Support Vector Machine (SVM) - Support vector machine (SVM) [30] is a supervised machine
learning algorithm that creates a hyperplane to separate N features, by mapping these features
to a multidimensional space. Since PD voice data is not linearly separable, we use an SVM
kernel to transform data into higher dimensional space. SVM performs well for PD data due to
memory efficiency and support vectors formed from a subset of training data points.
19. K nearest neighbors(KNN)-K nearest neighbors [31] (KNN) is a non-parametric, supervised

machine learning algorithm that groups data into clusters based on underlying similarities. It
works best for balanced audio data of 109 records due to small dataset size. Two clusters for
PWP and healthy data are created in an efficient manner. KNN is a lazy learning algorithm,
implying no presumptions of data are applied, ensuring novel patterns are learnt from
training data.
20. Alatas Bilal, Moradi Shadi, Tapak Leili, Afshar Saeid.”Identification of Novel Non
invasive Diagnostics Biomarkers in the Parkinson's Diseases and Improving the
Disease Classification Using Support run Machine’’.
BioMed Research International, 2022•hindawi.com
The purpose of this research was to discover genes involved in PD development. Methods. In
this study, the microarray dataset (GSE22491) provided by GEO was used for further analysis.
The Limma package under R software was used to examine and assess gene expression and
identify DEGs. The DAVID online tool was used to accomplish GO enrichment analysis and
KEGG pathway for DEGs. Furthermore, the PPI network of these DEGs was depicted using the
STRING database and analyzed through the Cytoscape to identify hub genes. Support vector
machine
(SVM) classifiers were subsequently employed to predict the accuracy of genes. Result. The PPI
network consisted of 264 nodes as well as 502 edges generated using the DEGs recognized from
the Limma package under the R software.
Moreover, three genes were identified as hubs: GNB5, GNG11, and ELANE. By using 3-gene
combination, SVM found that prediction accuracy of 88% can be achieved. Conclusion.
According to the findings of the study, the 3 hub genes GNB5, GNG11, and ELANE may be
used as PD detection biomarkers. Moreover, the results obtained from SVM with high accuracy
can be considered as PD biomarkers in further investigations.
21. Nicoló G. Pozzi, Ioannis U. Isaias (2022), “Chapter 19 - Adaptive deep brain
stimulation: Retuning Parkinson's disease”, Elsevier 184: 273-284.
https://doi.org/10.1016/B978-0-12-819410-2.00015-1
A brain-machine interface represents a promising therapeutic avenue for the treatment of many
neurologic conditions. Deep brain stimulation (DBS) is an invasive, neuro-modulatorytoolthat
can improve different neurological disorders by delivering electric stimulation to selected
brain areas. DBS is particularly successful in advanced Parkinson's disease (PD),where it allows
sustained improvement of motor symptoms. However, this approach is still poorly standardized,
with variable clinical outcomes.
22. Mei Jie, Desrosiers Christian, Frasnelli Johannes Machine Learning for the Diagnosis
of Parkinson's Disease: A Review of Literature
Frontiers in Aging Neuroscience, 13 (2021), 10.3389/fnagi.2021.633752
Diagnosis of Parkinson's disease (PD) is commonly based on medical observations and
assessment of clinical signs, including the characterization of a variety of motor symptoms.
However, traditional diagnostic approaches may suffer from subjectivity as they rely on the
evaluation of movements that are sometimes subtle to human eyes and therefore difficult to
classify, leading to possible misclassification.
In 3 studies, data from public repositories were combined with data from local databases or
participants (Agarwal et al., 2016; Choi et al., 2017; Taylor and Fenner, 2017). In the remaining
studies, data were sourced (Wahid et al., 2015) from another study (Fernandez et al., 2013),
collected at another institution (Segovia et al., 2019), obtained from the authors' institutional
database (Nunes et al., 2019), collected postmortem (Lewitt et al., 2013), or commercially
sourced (Váradi et al., 2019).
23. . Huang, C.X. Ling[20]Using AUC and accuracy in evaluating learning algorithms
The area under the ROC (receiver operating characteristics) curve, or simply AUC, has been
traditionally used in medical diagnosis since the 1970s. It has recently been proposed as an alternative
single-number measure for evaluating the predictive ability of learning algorithms. However, no
formal arguments were given as to why AUC should be preferred over accuracy. We establish formal
criteria for comparing two different measures for learning algorithms and we show theoretically and
empirically that AUC is a better measure (defined The efficiency of any content-based image retrieval
system depends on the extracted feature vectors of individual images stored in the database.
Generation of a compact feature vector with good discriminative power is a real challenge in the
image retrieval system. This paper presents the experimentation carried out to generate compact
feature vectors for a color image retrieval system based on image content.
25. Aghav-Palwe, S., Mishra, D. (2019)[11] “Color Image Retrieval Using Statistically Compacted
Features of DFT Transformed Color Images. In: Bhatia, S., Tiwari, S., Mishra, K., Trivedi, M. (eds)
Advances in Computer Communication and Computational Sciences. For dimensionality reduction
statistical parameters such as kurtosis, standard deviation, and variance are used for feature vector
generation. Euclidean distance is used in the proposed approach. Four different types of feature
vectors are created and tested for each image class.
CHAPTER 5 : PROPOSED METHEDOLOGY
5.1 DATA FLOW DIAGRAM (Level 0)

A Level 0 Data Flow Diagram (DFD) provides an initial high-level overview of the main
processes and interactions within a project system. It represents the system as a single
process and its connections to external entities. Level 0 DFD is also known as the Context
Diagram, as it defines the context within which the project operates.
Single Process: The Level 0 DFD depicts the entire project system as a single process. It
simplifies the complexity by treating the entire system as a black box.
External Entities: It shows the external entities that interact with the project system,
such as users, clients, or external systems. These entities exchange data with the system.
Data Flows: Data flows illustrate the movement of data between the external entities and
the central process. These data flows represent the inputs and outputs of the system.
High-Level View: Level 0 DFD provides a high-level view, focusing on the interactions
between the system and its external entities. It abstracts the internal details of the system.
Communication Tool: It serves as a communication tool among stakeholders, helping
them understand the scope and boundaries of the project.
figure 5.1: Level 0 DFD
5.2 DATA FLOW DIAGRAM (Level 1)

A Level 1 Data Flow Diagram (DFD) provides an overview of the main processes
involved in a system and the interactions between them, often focusing on the major
functions or processes within the system. Here's a description of the Level 1 DFD for
your pet adoption website.
Figure 5.2: Level 1 DFD for ADMIN
Figure 5.2: Level 1 DFD for USER
5.3 ER DIAGRAM
An Entity-Relationship Diagram (ERD) is a visual representation used in database design
to model the relationships between different entities and their attributes. ERDs are an
effective way to visualize the structure of a database, showing how different pieces of
information are related to each other. Here are some key details about ER diagrams:
Entities: Entities are objects or concepts that have distinct attributes. In a database
context, entities typically correspond to tables in a relational database. Examples of
entities could include "User," "Product," "Order," etc.
Attributes: Attributes are properties or characteristics of an entity. Each entity has a set
of attributes that describe it. For example, a "User" entity might have attributes like
"Username," "Email," and "Name."
Relationships: Relationships define how different entities are connected or associated

with each other. These relationships can be one-to-one, one-to-many, or many-to-many.
Relationships capture the interactions and dependencies between entities.
Cardinality: Cardinality describes the number of instances of one entity that are
related to the number of instances of another entity. Cardinality is denoted using symbols
like "1," "0 or more," and "1 to many."
Primary Key: A primary key uniquely identifies each record (row) in an entity. It
ensures the uniqueness of each entry and is typically used to establish relationships
between entities.
Foreign Key: A foreign key is a field in one entity that refers to the primary key of
another entity. It's used to establish relationships between entities and maintain data
integrity
Figure 5.3: ER
5.4 FLOW CHART
Figure 5.4: Flow chart
5.5 STRUCTURE CHART
Figure 5.5: Structure chart
CHAPTER 6 : HARDWARE AND SOFTWARE REQUIREMENTS
6.1 HARDWARE REQUIREMENT

I. Database Server:
A relational database management system (SQLITE3) to store user profiles, course

records, registration information.
II. Storage Solutions:
Adequate storage space for images, and pdf content associated with user profiles
and educational resources.
6.2 SOFTWARE REQUIREMENTS

I. Operating System:
A server-grade operating system (i.e., Windows Server) for hosting the website and
managing server resources.
II. Web Server:
A web server software (e.g., Apache) to handle HTTP requests and serve web content
to users
III. Database Management System:
A database management system (e.g., Sqlite3, MySQL, PostgreSQL) to store and

manage structured data, including pet profiles, vaccination records, and user data.
IV. Programming Languages:
● Backend: Python, JavaScript or other suitable languages for server-

side scripting and application logic
● Frontend: HTML, CSS, JavaScript, Bootstrap for building the user
interface and interactivity.
I. Frameworks and Libraries:
● Use frontend frameworks like Bootstrap for building dynamic

and responsive user interfaces.
● Backend frameworks like Django for building server-side applications.
II. Integrated Development Environment (IDE):
● IDEs like Visual Studio Code, PyCharm for coding, debugging,

and project management
CHAPTER 7 : DATABASE REQUIREMENT
Inbuilt Database (Sqlite3): An open-source relational database management system

known for its performance and scalability.
Database Design:
Designing the database schema to accommodate the various data entities within your
website. Key components include:
User Profiles: Design tables to store information about users and students and instructors,
including their attributes such as name, user id, password.
Data Integrity:
Implementing data integrity measures to ensure accuracy and consistency within the
database. This includes:
a. Primary Keys: Assign unique primary keys to each table to ensure that each
record can be uniquely identified.
b. Foreign Keys: Use foreign keys to establish relationships between
different tables, ensuring data consistency and referential integrity.
c. Constraints: Apply constraints such as NOT NULL, UNIQUE, and
CHECK constraints to maintain data quality.
Scalability and Performance:
Consider database optimization techniques to ensure your system can handle

increasing data loads.
Security:
Implement security measures to protect sensitive data, including proper access control and
encryption mechanisms.
Conclusion:
The database is the foundation of your data-driven Virtual Tutorial Management System.
Selecting the right DBMS, designing a well-structured schema, and implementing data
integrity and security measures are essential for creating a reliable and efficient system.
CHAPTER 8 : TECHNOLOGY USED
HTML (HyperText Markup Language):
HTML provides the structure for web pages in the project. It defines the layout and
organization of content.
In the context of a Parkinson's detection system, HTML would be used to create the user
interface elements such as forms, buttons, input fields, and display areas where users can
interact with the application.
CSS (Cascading Style Sheets):
CSS is used to style the HTML elements, defining their appearance, layout, and visual
presentation.
In the Parkinson's detection system, CSS would be utilized to make the user interface
visually appealing and intuitive. This includes setting colors, fonts, spacing, and overall
design aesthetics to enhance user experience.
JavaScript:
JavaScript adds interactivity and dynamic behavior to web pages. It allows for client-side
scripting, enabling features like form validation, user input handling, and real-time
updates without reloading the page.
In the Parkinson's detection system, JavaScript can be employed for tasks such as client-
side validation of input data, interactive visualizations of diagnostic results, and
asynchronous communication with the server for improved responsiveness.
Bootstrap:
Bootstrap is a popular CSS framework that provides pre-designed templates, components,
and responsive grid layouts for building modern web applications.
In the project, Bootstrap can be used to streamline the development process by leveraging
its pre-built UI components and responsive design features. This ensures that the
application is mobile-friendly and accessible across different devices and screen sizes.
BACKEND TECHNOLOGIES
Python:
Python is a versatile programming language that can be used for various tasks in the
Parkinson's detection system project, particularly on the server-side.
Python can be employed for backend development, handling data processing, analysis,
and model inference. It can interact with databases, perform machine learning tasks for
disease detection using libraries like scikit-learn or TensorFlow, and serve predictions to
the frontend via APIs.
Additionally, Python frameworks such as Flask or Django can be utilized to build the
backend server that communicates with the frontend, manages user authentication, and
serves the machine learning models for prediction.
CHAPTER 9 : IMPLEMENTATION
Methodology
This section explains the steps taken to achieve the prediction of Parkinson’s disease using
various machine learning. The various steps taken are Data gathering , Data Preprocessing,
Model Selection, Training, Evaluation, prediction .
Data Gathering
The first step is Data gathering .This step is very important because the quality and quantity of
the data you gather will directly affect the level of your prediction model. So we have taken data
of different Drawing recordings of the patient.
Data preparation
In this step the data is visualized well to spot the relationship between the parameters present in
the data so as to take the advantage of as well as to get the data imbalances. With this ,we need
to split the data into two parts .The first part for training the model like in our model we have
used 70 percent of data for training and 30 percent for testing. Which is the second part of the
data
Model Selection
The next step in our workflow is model selection. There are various models that have been used
till date by researchers and scientists. Some are meant for image processing ,some for sequences
like text, numbers or patterns. In our case we have 26 features which define the Drawing
recording of various patients so we have chosen such models which will classify or
differentiates the unhealthy patient with the healthy one.
Training
Training the dataset is one of the main tasks of machine learning .We will apply the data to
progressively improve the selected model’s ability to predict better i.e. the actual result should be
approx. to predict one.
Evaluation
The metrics we have calculated are ROC, Accuracy , Specificity , Precision etc. which will
highlight the best algorithm among all.
Prediction
In this phase we finally get the model ready to detect the prediction of Parkinson’s disease based
on the given dataset.
4.3 Using Python Tool on Standalone machine Environment

The Python computer programs are an essential tool for progression in the numeric examination
and machine learning spaces. Python is a perfect way to deal with reproducible, extraordinary
examination. Python is extensible and offers rich value for architects to manufacture their own
specific gadgets and procedures for examining data. With machines winding up recognizably
more basic as data generators, the noticeable quality of the dialects must be depended upon to
create.
In this module, the accuracy of different machine learning algorithms has been explored using
the Python Tool on the Standalone machine. Here initial analysis has been done using Jupyter
Notebook. A csv file has been provided as an input for Python Tool. Analysis has been done
using programming language Python as illustrated in fig 9.
Fig 9: Workflow of training the models of ML in Python
In any case, R has both upsides and downsides that designers ought to know. With enthusiasm
for the programming developing, as appeared on language notoriety files, for example, TIobe,
Redmond and PyPL, R initially showed up in the 1990s and has filled in as an execution of the S
measurable programming languages.
"Python is the most mainstream dialect utilized as a part of the field of statistics."It has all the
adaptability and power. Python is in reality only accumulations of scripts that are sorted out into
projects."
Data purifying/cleaning is a term identified with getting the significant data from the crude
information and noisy data removal (information not profitable to us). This should be possible
effectively in Jupyter Notebook and is a generally utilized strategy for each information
researcher.
Evaluation Criteria Used for Classification
Performance evaluations measures are the parameters which help in comparative analysis of
different machine learning techniques i.e. it tells the best algorithm among all other algorithms or
methods which can be used by medical science in the early prediction of neurodegenerative
diseases.
We have used several measures to evaluate the predictive results. These measures are average
absolute error (AAE), average related error (ARE), accuracy (ACC), Precision, Receiver
Operating Characteristics (ROC) , Area under ROC curve (AUC) ,sensitivity and specificity.
Let’s understand the performance evaluation measures.
Correlation Matrix
The confusion matrix is also called the Error matrix. It is a table that is often used to describe the
performance of a classification method on a set of test data for which actual values are known.
Each column of the matrix represents the instances in a predicted class. the correlation matrix is
represented as given
Predicted
No Yes
No TN FP
Actual
Yes FN TP
True Positive: is the count of healthy patients predicted accurately as healthy

True Negative: is the count of diseased subjects accurately predicted diseased.
False Positive: is the count of diseased patients predicted as healthy
False Negative: is the count of healthy patients predicted to be diseased
Accuracy and Precision

In classification, accuracy and precision are two important evaluation parameters. Accuracy is
the proportion of the total number of predictions that were correct. It can be obtained by the sum
of true positive and true negative instances divided by 100.And Precision is the fraction of true
positive and predicted yes instances. The formula of Accuracy and Precision are given below
Recall and F-Square

Recall is defined as the fraction between True Positive instances and Actual yes instances whereas
F-Square is the fraction between product of the recall and precision to the summation of recall and
precision parameter of classification. The formula of recall and precision given below
Sensitivity, Specificity and ROC

Sensitivity is defined as the fraction of true positive and actual yes instances whereas specificity is
the difference between one and false positive rate value. .ROC is defined as the fraction between
true positive rate and false positive rate.
CHAPTER 10: RESULTS
We have explained the implementation of different machine learning techniques and the
feature selection method i.e. the Proposed method. We have divided our work in two parts:
1. In the first part, we have run seven machine learning models and evaluated their error rates
i.e. AAE and ARE as given in table 4.
2. After that, based on their different error rates, we have evaluated and compared all the
methods on the basis of their accuracy, ROC, sensitivity, specificity and precision values
3. We found that Convolutional Neural Networks is the most efficient algorithm with the
accuracy of 87%, Precision 85.0%, ROC 96.4%.
4. In the second part , we are trying to select the most important and minimum number of
features from the Spiral Test data of 31 people where we have 23 features as explained in
chapter 4 in the dataset description .
5. For that we have used Feature selection whose working is shown in fig 12 by changing the
number of features selected in multiples of 5 i.e. firstly we check over 20 features then 15
features, 10 features and lastly 5 features.
6. From all the experiments Convolutional Neural Networks with 20 features selection
outstands from all the other ML techniques as it is giving the overall accuracy 96.6%, ROC
value 93.6 and precision of
88.7 which is better from all other machine learning techniques when compared with 5,10
and 15 feature’s performance metrics.
Fig 10: Feature selection by Proposed method
AAE and ARE
1. Average absolute error (AAE): AAE measure is calculated by taking the difference of
predicted value and the actual value .It can be understand average absolute error by
the below equation
n
AAE = (1/n) ∑ │( Ῡi – Y )│............................................................................................5
I=1
In eq 5, Ῡi is the predicted status value of the patient.

Y is the corresponding actual status value of the patient. .
n is the total number of parameters/columns.
2. Average related error (ARE): ARE measures how large the absolute error is
compared with the total size of the object measured. It is defined by the below
equation
n
ARE = (1/n) ∑ │( Ῡi – Y )│ /( Yi+1 )............................................................................6
I=1
In eq 6, Ῡi = predicted status value of the patient.
Y = actual status value of the patient.
n = total number of parameters/columns.
In Average related error, sometimes the value of Yi can be zero, to make the definition
we defined , we need to add ‘1’ with the value of Yi at the denominator . A small value
of AAE and ARE measures indicates that we have a good classification model. The
calculated values are shown in Table 4.
Table 1: Error rates analysis of seven classification methods
Sno. Classifier AAE ARE
1. Adaboost 0.040609 0.035533
2. DNN 0.055838 0.048223
3. Decision tree 0.060914 0.055838
4. Deep Neural Networks 0.274112 0.187817
5. Convolutional Neural 0.040609 0.035533

Networks
6. Support vector machine 0.045685 0.045685
7. Linear Regression 0.050761 0.043147
From the above analysis, we can conclude that Convolutional Neural Networks is the better
model as we know less the error rates, more efficient is the algorithm.
1.1 AAE and ARE analysis
AAE and ARE analysis as given in table 4 that Convolutional Neural Networks and Adaboost
techniques come up with the lower error rate values and hence outstands best from all the other
techniques used. Support vector machines and RNN are better classification techniques. While
Deep Neural Networks and decision trees have produced relatively lower accuracy as AAE and
ARE values are higher as compared to the other techniques.
In Fig 11 and 12, X-axis shows the prediction techniques and Y-axis shows the values of error
produced by the classification techniques.
Comparative Analysis of classification techniques.
We have performed a comparative analysis of seven classification techniques and we have

calculated the performance metrics such as accuracy, ROC, Precision, Confusion Matrix,
Sensitivity, Specificity. Based on which we have predicted the best machine learning technique It
is observed that Convolutional Neural Networks is better in comparison to all other ML
techniques with accuracy.
87%, Precision 85 % , ROC 96.4% that can be see through the shaded table 5. After the
comparative analysis of ML techniques, each performance metrics has been observed and
the major metrics i.e. accuracy and ROC (receiver operating characteristic curve) has been
analyzed individually. Fig 14 shows the Accuracy analysis graph.
Accuracy analysis
With respect to accuracy, Convolutional Neural Networks shows better results when compared to
other techniques which can be seen by fig 13. After Convolutional Neural Networks, Adaptive
boosting outstands from support vector machine and linear regression. RNN has shown better
results than decision tree and worst results were shown by Deep Neural Networks.
ROC analysis
AUC stands for “Area under the ROC curve” i.e. it measures the performance across all possible
classifications thresholds values or we can say it is the probability that the model identifies the
random positive example more highly than random negative example. It is used for visualize
binary classifier performance ie with two output classes. With respect to the fig 14,
ConvolutionalNeural Networks outstands when compared to other methods with ROC curve
value of 97%.
Feature selection
So far, in this work we have found the accuracies of different models on all the 23 features
extracted by Patients. Also, we have selected the minimum features which alone can predict the
PD. Here, the Feature selection technique has been used to select the 5 number of features, 10
number of features, 15 number of features and 20 number of features among all the features.
After that performance of various machine learning techniques are evaluated with different
features selected by Proposed method as illustrated in table 6,7,8 and
Table 3: 20 Features selected by Proposed
S.N CLASSIFIERS ACCURACY ROC PRECISION

o.
1 AdaBoost 91.3 92.6 91.33
2 RNN 81.36 77.2 88.1
3. Deep Neural 77.3 66.4 75.8
Networks
4. Decision Tree 80.9 75.3 82.2
5. Convolutional 96.6 93.6 88.7

Neural
Networks
6. Support 83.05 82.7 81.1
vector
machine
7. Linear 81.36 82 87.2
Regressio
n
S.No.
CLASSIFIERS ACCURACY ROC PRECISION

1 AdaBoost 93.9 90.2 91
2 RNN 89 85.3 85
3. Deep Neural 81.3 75 80.1
Networks
4. Decision Tree 77.3 75.3 80.1
5. Convolutional 94.9 94.9 94.2

Neural Networks
6. Support 83.5 85.3 82
vector
machine
7. Linear Regression 84.7 94.7 91.5
S.No CLASSIFIERS ACCURACY ROC PRECISION

1 AdaBoost 93.2 92 91.6
2 RNN 86.44 87.1 92.7
3. Deep Neural Networks 60.4 89 85.3

4. Decision Tree 58.3 77.8 78
5. Convolutional Neural 91.53 97 91.1

Networks
6. Support 86.4 93.3 88.2
vector
machine
7. Linear Regression 89.8 93.8 90
S. CLASSIFIERS ACCURACY ROC PRECISION
No
1 AdaBoost 91.2 82 89
2 RNN 86.44 76.2 89.4
3. Deep Neural Networks 77.9 78 75.5
4. Decision Tree 60.4 72 63.2
5. Convolutional Neural 88.14 94.6 86.4

Networks
6. Support 87 92.8 88.2
vector
machine
7. Linear Regression 81.36 90.1 80.9
Fig 15: Accuracy versus No. of selected features.
It is observed that the Convolutional Neural Networks with 20 numbers of features selected by
feature selection algorithm provided the overall accuracy 96.6%, ROC value 93.6 and precision
of 88.7 techniques has shows best results when compared with 5,10 and 15 feature’s
performance metrics as shown in fig 15.The graph shows the accuracy versus feature selection
plot, where y axis represents the number of features selected by feature selection algorithm and
x axis represents the hit value of accuracy.
CHAPTER 11 :CONCLUSION AND FUTURE SCOPE
5.1 Conclusion
In this work, various prediction models for Parkinson’s disease detection. For this purpose seven
machine learning techniques i.e. are used such as adaptive boosting, RNN, Deep Neural
Networks, Convolutional Neural Networks, decision tree, SVM and linear regression. To obtain
the desired results, error rates are calculated i.e. AAE and ARE as well as four performance
metrics are evaluated. These four metrics are accuracy, sensitivity, ROC, specificity.
From the results, Convolutional Neural Networks stands out from all the other ML techniques
with the accuracy of 97%, Precision 85.0%, ROC 96.4%. After that , we tried to selected the
most important and minimum number of features from the Spiral Test data of 31 people where
we have 23 features as explained in chapter 4 in dataset description .For that we have used
Feature selection whose working is shown in fig 12 by changing the number of features selected
in multiples of 5 ie firstly we check over 20 features than 15 features, 10 features and lastly 5
features. From all the experiments Convolutional Neural Networks with 20 features selection
outstands from all the other ML techniques as it is giving the overall accuracy 96.6%, ROC
value 93.6 and precision of
88.7 which is better in comparison to all other machine learning techniques when compared with
5,10 and 15 feature’s performance metrics.
5.2 Future scope
In this study we have used machine learning techniques, however very little research has been
done on deep learning methods. In future, the work can be extended by using autoencoders to
reduce the number of features and to extract the most important from them. Also the dataset
used in this work is not so complex , so the autoencoder did not learn well from that but with a
complex dataset it would definitely give better results.
Appendix :A
List of figures
figure no figure name page no
fig 1 Structure of neuron 1
fig 2 Architecture of proposed 7

system
fig 3 sample dataset of 9

biomedical drawing
measurement of 31 people
fig 4 a single input neuron 9
fig 5 multilayer perceptron 10
fig 6 straight line plot in 11

regression
fig 7 Prediction process taken by 12

convolutional neural
network
fig 8 Representation of decision 13

tree
fig 9 Hyper plane classifying two 14

classes
fig 10 Level 0 DFD 24
fig 13 ER Diagram 27
fig 14 Flow Chart 28
fig 15 Structure chart 29
fig 16 Workflow 37
fig 17 Correlation Matrix 38
fig 18 Feature Selection 41
fig 19 Accuracy versus no of 52

selected Features
Appendix :B
List Of Tables
Table No Table Title Page no
Table 1 Errors rates analysis of seven classification method 43
Table 2 Comparison between all the ML techniques using 44

performance matrix
Table 3 feature selected by proposed method 46
Table 4 feature selected by proposed method
APPENDIX:C
app.py
from flask import *

import os
from werkzeug.utils import secure_filename
import label_image
def load_image(image):
text = label_image.main(image)
return text
app = Flask( name )
@app.route('/')
@app.route('/first')
def first():
return render_template('first.html')
@app.route('/login')
def login():
return render_template('login.html')
@app.route('/chart')
def chart():
return render_template('chart.html')
@app.route('/index')
def index():
return render_template('index.html')
@app.route('/predict', methods=['GET',
'POST']) def upload():
if request.method == 'POST':
# Get the file from post
request f = request.files['file']
file_path = secure_filename(f.filename)
f.save(file_path)
# Make
prediction
result = load_image(file_path)
result = result.title()
d = {"brownspot":" → You are Normal. Keep Calm and Make Healthy Life",
'Normal':" → You are Normal. Keep Calm and Make Healthy Life",
"Severe":" → Neurodegenerative diseases are incurable and debilitating conditions that result in
progressive degeneration and / or death of nerve cells. This causes problems with movement (called ataxias),
mental functioning (called dementias) and affect a person's ability to move, speak and breathe[1].
Neurodegenerative disorders impact many families - these disorders are not easy for the individual nor their
loved ones.",
"Stage1":" → The most common neurodegenerative disorders are amyloidoses, tauopathies, α-
synucleinopathies, and TDP-43 proteinopathies. The protein abnormalities in these disorders have abnormal
conformational properties.",
"Stage2":" → Neurodegenerative diseases are often presented as a distinct entity, however there is often
overlap as you may have noted in the above descriptions, eg for AD and Lewy body pathologies. None of the
neurodegenerative disorders have perfect diagnostic accuracy, and neuropathology will continue to be the
gold standard for the foreseeable future."}
result = result+d[result]
#result2 =
result+d[result] #result =
[result]
#result3 = d[result]
print(result)
#print(result3)
os.remove(file_path)
return result
#return result3
return None
if name == ' main ':

app.run()
label_image.py
from future import

absolute_import fromfuture import
division
from future import print_function
import argparse
import sys
import time
import numpy as np
import tensorflow.compat.v1 as tf
tf.disable_v2_behavior()
def load_graph(model_file):
graph = tf.Graph()
graph_def = tf.compat.v1.GraphDef()
with open(model_file, "rb") as f:

graph_def.ParseFromString(f.read())
with graph.as_default():
tf.import_graph_def(graph_def)
return graph
def read_tensor_from_image_file(file_name, input_height=299, input_width=299,

input_mean=0, input_std=255):
input_name = "file_reader"
output_name =
"normalized"
file_reader = tf.io.read_file(file_name,
input_name) if file_name.endswith(".png"):
image_reader = tf.image.decode_png(file_reader, channels = 3,
name='png_reader')
elif file_name.endswith(".gif"):
image_reader = tf.squeeze(tf.image.decode_gif(file_reader,
name='gif_reader'))
elif file_name.endswith(".bmp"):
image_reader = tf.image.decode_bmp(file_reader,
name='bmp_reader') else:
image_reader = tf.image.decode_jpeg(file_reader, channels = 3,
name='jpeg_reader')
float_caster = tf.cast(image_reader, tf.float32)
dims_expander = tf.expand_dims(float_caster, 0);
resized = tf.image.resize(dims_expander, [input_height,

input_width]) normalized = tf.divide(tf.subtract(resized,
[input_mean]), [input_std]) sess = tf.Session()
result =
sess.run(normalized) return
result
def load_labels(label_file):
label = []
proto_as_ascii_lines = tf.gfile.GFile(label_file).readlines()
for l in proto_as_ascii_lines:
label.append(l.rstrip())
return label
def main(img):
file_name = img
model_file =
"retrained_graph.pb" label_file =
"retrained_labels.txt"
input_height = 299
input_width = 299
input_mean = 128
input_std = 128
input_layer = "Mul"
output_layer = "final_result"
parser = argparse.ArgumentParser()
parser.add_argument("--image", help="image to be processed")
parser.add_argument("--graph", help="graph/model to be executed")
parser.add_argument("--labels", help="name of file containing
labels") parser.add_argument("--input_height", type=int, help="input
height") parser.add_argument("--input_width", type=int, help="input
width") parser.add_argument("--input_mean", type=int, help="input
mean") parser.add_argument("--input_std", type=int, help="input
std") parser.add_argument("--input_layer", help="name of input
layer") parser.add_argument("--output_layer", help="name of output
layer") args = parser.parse_args()
if args.graph:
model_file = args.graph
if args.image:
file_name = args.image
if args.labels:
label_file = args.labels
if args.input_height:
input_height = args.input_height
if args.input_width:
input_width =
args.input_width if
args.input_mean:
input_mean =
args.input_mean if
args.input_std:
input_std =
args.input_std if
args.input_layer:
input_layer = args.input_layer
if args.output_layer:
output_layer = args.output_layer
graph = load_graph(model_file)
t = read_tensor_from_image_file(file_name,
input_height=input_height,
input_width=input_width,
input_mean=input_mean,
input_std=input_std)
input_name = "import/" + input_layer

output_name = "import/" + output_layer
input_operation = graph.get_operation_by_name(input_name);
output_operation = graph.get_operation_by_name(output_name);
with tf.Session(graph=graph) as sess:

start = time.time()
results = sess.run(output_operation.outputs[0],
{input_operation.outputs[0]: t})
end=time.time()
results = np.squeeze(results)
top_k = results.argsort()[-5:][::-1]
labels = load_labels(label_file)
for i in top_k:
return labels[i]
index.html
{% extends "base.html" %} {% block content %}
<center><h2>Disease Classifier</h2>
<style> .footer {
position: fixed;
left: 0;
bottom: 0;
width: 100%;
background-color: #2dcba7;
color: black;
text-align: center;
font-size: 25px;
font-weight: bold;
}</style>
<div>
<center> <form id="upload-file" method="post" enctype="multipart/form-data">
<label for="imageUpload" class="upload-label">
Choose Image...
</label>
<center> <input type="file" name="file" id="imageUpload" accept=".png, .jpg, .jpeg">
</form>
<div class="image-section" style="display:none;">

<div class="img-preview">
<div id="imagePreview">
</div>
</div>
<div>
<button type="button" class="btn btn-primary btn-lg " id="btn-predict">Analyze Spiral</button>
</div>
</div>
<div class="loader" style="display:none;"></div>
<h3 id="result">
<span> </span>
</h3>
</div>
<h3 id="result3">
}
</script>
</head>
<body>
<div id="chartContainer" style="height: 370px; width: 100%;"></div>
<script src="https://canvasjs.com/assets/script/canvasjs.min.js"></script>
</body>
</html>
{% endblock %}
login.html
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta content="width=device-width, initial-scale=1.0" name="viewport">
<title>Neurodegenerative Disorder</title>
<meta content="" name="description">
<meta content="" name="keywords">


<link href="../static/img/favicon.png" rel="icon">
<link href="../static/img/apple-touch-icon.png" rel="apple-touch-icon">


<link href="https://fonts.googleapis.com/css?family=Open+Sans:300,300i,400,400i,600,600i,700,700i|
Playfair+Di splay:ital,wght@0,400;0,500;0,600;0,700;1,400;1,500;1,600;1,700|
Poppins:300,300i,400,400i,500,500i,600,
600i,700,700i" rel="stylesheet">


<link href="../static/vendor/animate.css/animate.min.css" rel="stylesheet">
<link href="../static/vendor/aos/aos.css" rel="stylesheet">
<link href="../static/vendor/bootstrap/css/bootstrap.min.css" rel="stylesheet">
<link href="../static/vendor/bootstrap-icons/bootstrap-icons.css" rel="stylesheet">
<link href="../static/vendor/boxicons/css/boxicons.min.css" rel="stylesheet">
<link href="../static/vendor/glightbox/css/glightbox.min.css" rel="stylesheet">
<link href="../static/vendor/swiper/swiper-bundle.min.css" rel="stylesheet">


<link href="../static/style.css" rel="stylesheet">

</head>
<body>



<header id="header" class="fixed-top d-flex align-items-cente">
<div class="container-fluid container-xl d-flex align-items-center justify-content-lg-between">
<h1 class="col-lg-8"><a href="index.html"> <span>Neurodegenerative Disorder Disease

Classification</span></a></h1>


<nav id="navbar" class="navbar order-last order-lg-0">

<ul>
<li><a class="nav-link scrollto " href="{{ url_for('first')}}">Home</a></li>
<li><a class="nav-link scrollto active" href="{{ url_for('login')}}">Login</a></li>
</ul>
</div>
</header>


<section id="hero" class="d-flex align-items-center">
<div class="container position-relative text-center text-lg-start" data-aos="zoom-in" data-aos-
delay="100">
<div class="row">
<div class="col-lg-8">
<h1><span>Neurodegenerative Disorder Disease Prediction Using RCNN Classification</span></h1>
</div>
</div>
</div>
</section>
<section id="book-a-table" class="book-a-table">

<div class="container" data-aos="fade-up">
<head>
<script>
addEventListener("load", function () {
setTimeout(hideURLbar, 0);
}, false);
function hideURLbar() {
window.scrollTo(0, 1);
}
function login(){
var uname =
document.getElementById("uname").value; var pwd =
document.getElementById("pwd").value;
if(uname == "admin" && pwd == "admin")

{
alert("Login Success!");
window.location = "{{url_for('index')}}";
return false;
}
else
{
alert("Invalid Credentials!")
}
}
</script>
</head>
<body id="page-top">


<section class="page-section portfolio" id="portfolio">
<br>
<br>


<div class="section-title">
<center> <p>Login</p></center>
</div>

<div class="row">


<div class="col-md-6 col-lg-4" style="margin-left:380px">
<div class="control-group">

<label class="control-label" for="username"><b>Username</b></label>

<div class="controls">
<input type="text" id="uname" name="uname" placeholder="" class="form-
control">
</div>
</div>

</br>
<label class="control-label" for="password"><b>Password</b></label>
<input type="password" id="pwd" name="pwd" placeholder="" class="form-
control">
</div>
</div>
<div class="col-md-6 col-lg-4" style="margin-left:-150px">

<br>
<input type="button" class="btn btn-success" value="Login" style="margin-

left: 280px" onclick="login()">
</div>
</div>
</div>
</div>
</div>
</section>
</body>
</div>
</section>








<footer id="footer">
<div class="container">
<div class="copyright">
</div>
<div class="credits">




</div>
</div>
</footer>
<div id="preloader"></div>
<a href="#" class="back-to-top d-flex align-items-center justify-content-center"><i class="bi bi-
arrow- up-short"></i></a>


<script src="../static/vendor/aos/aos.js"></script>
<script src="../static/vendor/bootstrap/js/bootstrap.bundle.min.js"></script>
<script src="../static/vendor/glightbox/js/glightbox.min.js"></script>
<script src="../static/vendor/isotope-layout/isotope.pkgd.min.js"></script>
<script src="../static/vendor/php-email-form/validate.js"></script>
<script src="../static/vendor/swiper/swiper-bundle.min.js"></script>


<script src="../static/js/main.js"></script>
</body>
</html>
APPENDIX:D
REFERENCES
[1] Kamal Nayan Reddy, Challa, Venkata Sasank Pagolu and Ganapati Panda, “An Improved
Approach for Prediction of Parkinson’s Disease using Machine Learning Techniques”, in
Procedings of the International conference on Signal Processing, Communication, Power and
Embedded System (SCOPES)-2016, pp. 1446-145, 2016.
[2] Geeta Yadav, Yugal Kumar and G. Sahoo, “Predication of Parkinson’s disease using Data
Mining Methods: a comparative analysis of tree, statistical and support vector machine
classifiers”, in Procedings of the National Conference on Computing and Communication
Systems (NCCCS),
pp. 1-4, 2012.
[3] Paolo Bonato, Delsey M. Sherrill, David G. Standaert, Sara S. Salles and Metin Akay, “Data
Mining Techniques to Detect Motor Fluctuations in Parkinson's Disease”, in Proceedings of the
26th Annual International Conference of the IEEE Engineering in Medicine and Biology Society,
pp. 4766-4769, 2004.
[4] Sonu S. R., Vivek Prakash and Ravi Ranjan, “Prediction of Parkinson’s Disease using Data
Mining”, in Proceedings of the International Conference on Energy, Communication, Data
Analytics and Soft Computing (ICECDS), pp. 1082-1085, 2017.
[5] Aarushi Agarwal, Spriha Chandrayan and Sitanshu S Sahu, “Prediction of Parkinson’s
Disease using Speech Signal with Extreme Learning Machine”, in Proceedings of the
International Conference on Electrical, Electronics, and Optimization Techniques (ICEEOT), pp.
1-4, 2016.
[6] Akshaya Dinesh and Jennifer He, “Using Machine Learning to Diagnose Parkinson’s Disease
from Drawing Recording”, in Proceedings of the IEEE MIT Undergraduate Research
Technology Conference (URTC), pp. 1-4, 2017.
[7] Giulia Fiscon, Emanuel Weitschek, Giovanni Felici and Paola Bertolazzi, “Alzheimer’s disease
patients classification through EEG signals processing”, in Proceedings of the IEEE Symposium
on Computational Intelligence and Data Mining (CIDM). pp 1-4, 2014.
[8] Pedro Miguel Rodrigues, Diamantino Freitas and Joao Paulo Teixeirab, “Alzheimer
electroencephalogram temporal events detection by K-means”, in Proceedings of the
International Conference on Health and Social Care Information Systems and Technologies
HCIST. pp. 859 – 864, 2012.
[9] Elva Maria Novoa-del-Toro, Juan Fernandez-Ruiz, Hector Gabriel Acosta-Mesa and
Nicandro Cruz-Ramirez, “Applied Macine Learning to Identify Alzheimer's Disease through the
Analysis of Magnetic Resonance Imaging”, in Proceedings of the International Conference on
Computational Science and Computational Intelligence, pp. 577-582, 2015.
[10] Daniel Johnstone1, Elizabeth A. Milward1, Regina Berretta1 and Pablo Moscato1,
“Multivariate Protein Signatures of Pre-Clinical Alzheimer’s Disease in the Alzheimer’s Disease
Neuroimaging Initiative (ADNI) Plasma Proteome Dataset”, in Proceedings of the Disease
Neuroimaging Initiative, vol-7, pp. 1-17, 2017.
[11] Jason Orlosky, Yuta Itoh, Maud Ranchet, Kiyoshi Kiyokawa, John Morgan, and Hannes
Devos, “Emulation of Physician Tasks in Eye-tracked Virtual Reality for Remote Diagnosis of
Neurodegenerative Disease”, in Proceedings of the IEEE Transactions on Visualization and
Computer Graphics, vol. 23, pp. 1302 – 1311, 2017.
[12] Mathew J. Summers, Vienna, Austria, Alessandro E. Vercelli, Georg Aumayr, Doris M.
Bleier and Ludovico Ciferri,”Deep Machine Learning Application to the Detection of Preclinical
Neurodegenerative Diseases of Aging”, in Proceedings of the Scientific Journal on Digital
Cultures, vol. 2, pp. 9-24, 2017.
[13] Bianca Torres, Raquel Luiza Santos, Maria Fernanda Barroso de Sousa, Jose Pedro Simoes
Neto, Marcela Moreira Lima Nogueira, Tatiana T. Belfort1, Rachel Dias1, Marcia and Cristina
Nascimento Dourado, “Facial expression recognition in Alzheimer’s disease: a longitudinal
study”,
pp. 383-389, 2014.
[14] Smitha Sunil and Kumaran Nair, “An exploratory study on Big data processing: a case study
from a biomedical informatics”, 3rd MEC International Conference on Big Data and Smart City,
pp. 1-4, 2016.
[15] C. Kotsavasilogloua, N. Kostikis, D. Hristu-Varsakelis and M. Arnaoutoglouc, “Machine

learning-based classification of simple drawing movements in Parkinson’s disease”, in
Proceedings of the Biomedical Signal Processing and Control, pp. 174–180, 2017.
[16] Santosh S. Rathore and Sandeep Kumar, “An empirical study of some software fault
prediction techniques for the number of faults prediction”, in Proceedings of the Soft Computing,
vol. 21, pp 7417–7434, 2017.
[17] Arvind Kumar Tiwari, “Machine Learning Based Approaches for Prediction of Parkinson’s
Disease”, in Proceedings of the Machine Learning and Applications: An International Journal
(MLAIJ), vol.3, pp. 33-39, 2016.
[18] Polina Mamoshina, Armando Vieira, Evgeny Putin and Alex Zhavoronkov, “Applications of
Deep Learning in Biomedicine”, in Proceedings of the American Chemical Sociecty Mol.
Pharmaceutics, pp. 1445−1454, 2016.
[19] Alexis Elbaz,James H. Bower, Brett J. Peterson, Demetrius M. Maraganore, Shannon K.

McDonnell, J. Eric Ahlskog, Daniel J. Schaid, Walter A. Rocca, “Survival Study of Parkinson
Disease in Olmsted County, Minnesota “, Arch Neurol. Vol. 60 pp. 91-96, 2003.
[20] Tanner CM, Ross GW, Jewell SA ,” Occupation and risk of Parkinsonism: a multicentercase-
control study” Arch Neurol,66(9):1106–1113,200
[21] V. A. Sukhanov, I. D. Ionov, and L. A. Piruzyan, “Neurodegenerative Disorders: The Role of

Genetic Factors in Their Origin and the Efficiency of Treatment” in Proceedings of the Human
Physiology US National Library of Medicine National Institutes of Health, vol. 31, pp. 472–482,
2005.
[22] Marras C, Tanner C.”Epidemiology of Parkinson's Disease”, Movement Disorders:

Neurologic Principles and Practice, 2nd ed.2004, Watts, RL, Koller, WC (Eds). The McGraw-
Hill Companies:New York, pp. 177.
[23] http://www.orionpharma.co.uk/Products-and-Services-Orion/Parkinsons-disease/10-facts-
about-Parkinsonsdisease/
[24] Cnockaert, L., Schoentgen, J., Auzou, P., Ozsancak, C.,Defebvre, L., & Grenez, F., “Low
frequency vocal modulations in vowels produced by Parkinsonian subjects”, Speech
Communications, vol 50, pp. 288-300, 2008.
[25] Kenneth Revett, Florin Gorunescu and Abdel-Badeeh Mohamed Salem, “Feature Selection in
Parkinson’s disease: A Rough Sets Approach”, Proceedings of the International Multi onference
on Computer Science and Information Technology, pp. 425 – 428,2004, ISBN 978-83- 60810-
22-
4.

Final Documentation (Major Project)

Uploaded by

Copyright:

Available Formats

You might also like

Final Documentation (Major Project)

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Final Documentation (Major Project)

Uploaded by

Copyright:

Available Formats

A PROJECT REPORT

PARKINSON’S DETECTION SYSTEM

Submitted in the Partial Fulfillment of the Requirement for the Award of

Computer Science & Engineering (2024)

Astha Dixit [2000460100018]

Under the Guidance Of

Certified that the project entitled “PARKINSON’S DETECTION SYSTEM

Asst. Prof. (Mr.) Milind Bhatt Prof. (Dr.) Atul Kumar

(Project Guide) (Head of Dep

We hereby declare that the project entitled “PARKINSON’S DETECTION SYSTEM”

Astha Dixit Astha Mishra

Janhavi Srivastava Priti Yadav

Parkinson's disease is a complex neurodegenerative disorder that affects millions of people

Table of Contents xi-xii

4. Literature Review 16-23

5.1 Data Flow Diagram (Level 0)

5.2 Data Flow Diagram (Level 1)

5.4 Flow Chart

5.5 Structure Chart

6. Hardware and Software Requirements 31-32

6.1 Hardware Requirements

6.2 Software Requirements

8.2 Backend Technologies

10. Result 41-52

11. Conclusion and future scope of the project 53

Fig 1: Structure of neuron present in human brain

listed factor which caused Parkinson’s disease in an individual.

A. Environmental factors: Environment is defined as the surrounding or the place in which an

a. Exposure to heavy metals (like lead and aluminum) and pesticides.

Parkinson’s Disease Symptoms

The project objective of a Parkinson's detection system is to develop a technology or system

Remote Monitoring: Enable individuals with Parkinson's to be monitored remotely, reducing

Cost-Effectiveness: Consider the cost implications of the system, aiming to provide an

CHAPTER 3 : WORK PROPOSAL

Fig 2: Sample dataset of biomedical Drawing measurements of 31 people

Fig 3: A Single input neuron

Fig 4 : Multilayer perceptron

O = 𝛔(𝛏𝟏 𝛚𝟏 + 𝛏𝟐 𝛚𝟐 + ⋯ … … + 𝛏𝐤𝛚𝐤) +Θ...........................................................................2

o is the sigmoid function or transformed function

ω is the weight of input (1 to k)

Fig 5: Straight line plot in Linear regression

Y=b0 +b1X1 +b2X2..............................................................................................................................................4

In eq 3 and eq 4, Y is the dependent variable and X1,X2 represents the independent

Convolutional Neural Networks

Test sets features

Fig 6: Prediction process taken by Convolutional Neural Networks

Fig 7 : Representation of decision tree

Mathematics involved in Adaboost :

H(x) = sign( ∑ ah(x)

h(x) is the output of weak classifier t for input

Support Vector Machine

Fig 8: Hyper plane classifying two classes.

CHAPTER 4 : LITERATURE REVIEW

1. “Pathoanatomy of Parkinson’s disease” a book written by Braak, H., Braak, E. has

7. Wroge et. al [14] aimed to remove subjectivity of doctors in prediction of PD using an

19. K nearest neighbors(KNN)-K nearest neighbors [31] (KNN) is a non-parametric, supervised

CHAPTER 5 : PROPOSED METHEDOLOGY

5.1 DATA FLOW DIAGRAM (Level 0)

figure 5.1: Level 0 DFD