Shri Shamrao Patil (Yadravkar) Educational & Charitable Trust’s

Sharad Institute of Technology College of Engineering

(Autonomous INSTITUTE )Yadrav (ichalkarnji)-416121

(Approved by AICTE, New Delhi, Recognized by Government of Maharashtra

& Affiliated to DBATU)

DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING

A
Project Report
On

“Autism spectrum disorder prediction”

Submitted in partial fulfillment of the requirement for the degree of

Bachelor of Technology
In
COMPUTER SCIENCE & ENGINEERING

by

1. Ms. Harshada Sarjerao Pawar 2110104056

2. Ms. Mayuri Pandurang Ravan. 2110104012
3. Ms. Manasi Narasgonda Sangale. 2140104117
4. Ms. Benzeer Jahur Vathare. 2110104025

under the guidance of

Ms.F.A.Patel

Department of Computer Science & Engineering,

 Sharad Institute of Technology college of Engineering, Yadrav-Ichalkaranji
2023-24

Shri Shamrao Patil (Yadravkar) Educational & Charitable Trust’s

Sharad Institute of Technology College of Engineering
(Autonomous INSTITUTE )Yadrav (ichalkarnji)-416121
(Approved by AICTE, New Delhi, Recognized by Government of
Maharashtra & Affiliated to DBATU)
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING

This is to certify that,

Ms. Harshada Sarjerao Pawar. (2110104056)
Ms. Mayuri Pandurang Ravan. (2110104012)
Ms. Manasi Narasgonda Sangale. (2140104117)
Ms. Benzeer Jahur Vathare. (2110104025)

Studying in B-Tech. Computer Science & Engineering have successfully completed

the project entitled
“Autism spectrum disorder prediction”
under the guidance and supervision of Prof. Ms.K.M.Tamboli. during the academic
year 2020-2021 towards the partial fulfillment of curriculum prescribed by Dr.
Babasaheb Ambedkar Technological University, Lonere for the award of degree of
Bachelor of Technology in Computer Science & Engineering.

Ms.F.A.Patel Dr. S.B.Gurav

Guide Head of Department of
Dept. of Computer Science Computer Science& Engineering,
&Engineering, SITCOE, Yadrav
SITCOE, Yadrav

Dr. S A Khot
Principal External Examiner
SITCOE, Yadrav
 Autism Spectrum Disorder Detection

ACKNOWLEDGEMENT

We would like to express our deep and sincere gratitude to my Guide Prof.
Ms.F.A.Patel, Department of Computer Science & Engineering, for guiding us to
accomplish this project work. It was our privilege and pleasure to work under her able
guidance. We are indeed grateful to her for providing helpful suggestion, from time to
time. Due to her constant encouragement and inspiration we are able to present this
project.
We express our deep gratitude to Dr S.B.Gurav, Head of Computer Science
& Engineering Department, for his valuable guidance and constant encouragement.
We are very much thankful to Dr.S.A.Khot, Principal, Sharad institute of technology
college of engineering, Yadrav-Ichalkaranji for providing all the necessary facilities
to carry out project work.
Last but not least we are thankful to our parents for their moral as well as
financial support.

SITCOE,Yadrav
 Autism Spectrum Disorder Detection

ABSTRACT

Autism Spectrum Disorder (ASD) is a group of neurodevelopmental disabilities that

are not curable but may be ameliorated by early interventions. We gathered early-
detected ASD datasets relating to toddlers, children, adolescents and adults, and
applied several feature transformation methods, including log, Z-score and sine
functions to these datasets. Various classification techniques were then
implemented with these transformed ASD datasets and assessed for their
performance. We found SVM showed the best performance for the toddler dataset,
while Adaboost gave the best results for the children dataset, XGboost for the
adolescent and Adaboost for the adult datasets. The feature transformations
resulting in the best classifications was sine function for toddler and Z-score for
children and adolescent datasets. After these analyses, several feature selection
techniques were used with these Z-score-transformed datasets to identify the
signicant ASD risk factors for the toddler, child, adolescent and adult subjects. The
results of these analytical approaches indicate that, when appropriately optimized,
machine learning methods can provide good predictions of ASD status. This
suggests that it may possible to apply these models for the detection of ASD in its
early stages

SITCOE,Yadrav
 Autism Spectrum Disorder Detection

INDEX

List of Figures i
1 Introduction 1
1.1 Background 1
1.2 Motivation 1
1.3 Objective 2
1.4 Scope Of Work 2
2 Literature Review 3
3 Problem Statement 4
4 Objectives 5
5 Methodology 6
5.1 Methodology 7
5.2 Architecture 8
5.3 Data Flow Diagram 9
5.3.1 0 Level DFD 9
5.3.2 1 Level DFD 9
5.3.3 2 Level DFD 10
5.4 Use Case Diagram 10
5.5 Activity Diagram 11
6 Result and Discussion 12
6.1 Results 12
6.2 Screenshots 13
7 Future Work 16
8 Conclusion 17
9 References 18

SITCOE,Yadrav
 Autism Spectrum Disorder Detection

LIST OF FIGURES

Sr. No. Title Page No.

Figure 5.2.1 Architecture Diagram 8
Figure 5.3.1 0 Level DFD 9
Figure 5.3.2 1 Level DFD 9
Figure 5.3.3 2 Level DFD 10
Figure 5.4.1 Use Case Diagram 10
Figure 5.5.1 Activity Diagram 11

SITCOE, Yadrav Page i

 Autism Spectrum Disorder Detection

CHAPTER 1

INTRODUCTION
he problem of autism spectrum disorder (ASD) have been mounting swiftly nowadays
among all ages of the
human population. Early detection of this neurological disease can greatly assist in the
maintenance of the subject’s
mental and physical health. With the rise of application of machine learning-based
models in the predictions of various
he problem of autism spectrum disorder (ASD) have been mounting swiftly nowadays
among all ages of the
human population. Early detection of this neurological disease can greatly assist in the
maintenance of the subject’s
mental and physical health. With the rise of application of machine learning-based
models in the predictions of various
he problem of autism spectrum disorder (ASD) have been mounting swiftly nowadays
among all ages of the
human population. Early detection of this neurological disease can greatly assist in the
maintenance of the subject’s
mental and physical health. With the rise of application of machine learning-based
models in the predictions of various
he problem of autism spectrum disorder (ASD) have been mounting swiftly nowadays
among all ages of the
human population. Early detection of this neurological disease can greatly assist in the
maintenance of the subject’s
mental and physical health. With the rise of application of machine learning-based
models in the predictions of various
he problem of autism spectrum disorder (ASD) have been mounting swiftly nowadays
among all ages of the
human population. Early detection of this neurological disease can greatly assist in the
maintenance of the subject’s
mental and physical health. With the rise of application of machine learning-based
models in the predictions of v
1.1 Background
The background of a ASD is, Autism spectrum disorders (ASD) are a diverse
group of conditions. They are characterized by some degree of difficulty with
social interaction and communication. Other characteristics are atypical
patterns of activities and behaviours, such as difficulty with transition from
one activity to another, a focus on details and unusual reactions to sensations.

1.2 Motivation

SITCOE,Yadrav Page 1
 Autism Spectrum Disorder Detection

The motivation behind an Autism Spectrum Disorder (ASD) project typically

stems from the recognition of the challenges faced by individuals with ASD and
their families, as well as the desire to improve their quality of life.

There are several variables, for example, inheritance and surroundings, which
influence the growth of neurodevelopmental disorder such as autism spectrum
disorder (ASD) and attention deficit hyperactivity disorder (ADHD) during the
first 36 months of life. ADHD is the most predominant childhood disorder
initially characterized by behavior issues. In addition, ADHD is a multisystemic
neurodevelopmental condition which is linked to temporal, operational, and
structural connectivity impairments in the temporoparietal, front striatal, and
frontal cortex networks . ASD, like ADHD, is also initially characterized by
inappropriate behavior , including spatial memory difficulties .

The issues faced by children having ASD and ADHD are shown in Figure 1. Both
conditions appear at a young age and can last throughout adulthood, causing
multiple disabilities as well as significant burdens on individuals and families

Machine learning (ML) and deep learning (DL) are the subfields of artificial
intelligence (AI). Deep learning however has evolved faster in the last few years.
It is an advanced phase of Machine learning with a vast application from defense
to marketing. Deep learning presents healthcare with groundbreaking applications.
The primary strength of DL lies in gathering a massive amount of data. It applies
its advanced neural networks to come up with its best outcome.

Therefore, this study emphasizes on the use of ML techniques for the early
detection and treatment of symptoms related to ASD.

1.3 Objective

 To develop a prediction model that can accurately identify children’s and all aged
peoples at risk of autism spectrum disorder (ASD),

 To create a model to find signs of autism in kids early, helping families and
doctors get support sooner.

SITCOE,Yadrav Page 2
 Autism Spectrum Disorder Detection

 To put together different kinds of information like behaviour, genetics, and brain
scans to make the model smarter.

 To Make sure the model can give advice that fits each , so families know exactly
how to help.

1.4 Scope of work

2 The project is all about learning more about Autism Spectrum Disorder (ASD).
We want to understand what causes it, how we can tell if someone has it, what
other issues might come along with it, and what we can do to help. We'll look at
different ways to help people with ASD, like therapy or medicine, and also think
about what we still need to figure out. Ultimately, we want to make life better for
people with ASD and their families.

CHAPTER 2

LITERATURE REVIEW

Sr.no. Paper Name Author Year Description

1. "Neuroimaging Chen,H., The authors summarize recent

Studies in Patel, J. 2021 neuroimaging research in ASD,
Autism exploring structural and
Spectrum functional brain differences and
Disorder” their implications for
understanding the disorder's
neurobiology.

2. “Social Rodriguez, M., This paper examines the social

Communication Kim, S. 2022 communication difficulties
Challenges in experienced by individuals with
Autism ASD, discussing underlying

SITCOE,Yadrav Page 3
 Autism Spectrum Disorder Detection

Spectrum mechanisms and implications for

Disorder" social skill interventions and
support.

3. "Advances in Patel, R., This paper discusses recent

Early Detection Nguyen, L. 2023 advances in early detection
of Autism methods for ASD, including
Spectrum novel screening tools and
Disorder" technologies, and their
implications for improving
outcomes through early
intervention.

444. "Prediction of T. Lakshmi 2019 Autism appears to be a neuro

4444444. Autism Spectrum Praveena and developmental disorder that is
4. Disorder Using N. V. Muthu visible in the early years. It is a
Supervised Lakshmi
wide-spectrum disorder that
Machine
Learning indicates that the severity and
Algorithms”. symptoms can vary from person to
person.

CHAPTER 3

PROBLEM STATEMENT

To predict whether an individual is likely to have Autism Spectrum Disorder (ASD)

based on various features such as behavioral data, developmental milestones, genetic
information, and other relevant factors.

SITCOE,Yadrav Page 4
 Autism Spectrum Disorder Detection

CHAPTER 4

OBJECTIVES

1. Functional Independence: Enhance the individual's ability to perform daily

tasks independently, such as self-care, communication, and social interaction.

2. Skill Development: Focus on acquiring and improving essential skills,

including language and communication skills, motor skills, cognitive skills,
and adaptive behaviors.

3. Behavior Management: Develop strategies to address challenging behaviors

and promote positive behavior patterns, such as self-regulation, impulse
control, and social skills.

SITCOE,Yadrav Page 5
 Autism Spectrum Disorder Detection

4. Education and Learning: Support academic achievement by providing

appropriate educational interventions and accommodations tailored to the
individual's learning style and needs.

5. Social Integration: Facilitate opportunities for social interaction and inclusion

in various settings, such as school, community activities, and peer groups, to
foster social skills and friendships.

6. Quality of Life: Enhance the individual's overall quality of life by promoting

self-esteem, emotional well-being, and participation in meaningful activities
and relationships.

7. Health and Wellness: Address any medical or health-related issues associated

with the neurodevelopmental disorder, promote healthy lifestyle habits, and
ensure access to appropriate healthcare services.

CHAPTER 5

METHODOLOGY

5.1 Methodology :

Detecting autism spectrum disorder (ASD) using machine learning involves several
steps and considerations to ensure accuracy and reliability. Here's a methodology
outline:

1. Data Collection: Gather a diverse and representative dataset containing

features relevant to ASD diagnosis. This dataset may include clinical
observations, behavioral assessments, medical history, genetic information,
and demographic data. Ensure that the dataset complies with privacy
regulations and ethical standards.

SITCOE,Yadrav Page 6
 Autism Spectrum Disorder Detection

2. Feature Selection and Engineering: Identify relevant features that may

discriminate between individuals with ASD and neurotypical individuals. This
may involve domain knowledge, literature review, and statistical analysis.
Feature engineering techniques may also be applied to extract meaningful
information from raw data.
3. Data Preprocessing: Clean the dataset by handling missing values, outliers,
and inconsistencies. Normalize or standardize features to ensure uniformity
and facilitate model training. Split the dataset into training, validation, and test
sets to assess model performance.
4. Model Selection: Choose appropriate machine learning algorithms for
classification tasks. Commonly used algorithms for ASD detection include
logistic regression, support vector machines (SVM), decision trees, random
forests, and neural networks. Consider the trade-offs between interpretability,
complexity, and performance.
5. Model Training: Train the selected machine learning models using the
training dataset. Tune hyperparameters through techniques such as grid search
or random search to optimize model performance. Employ cross-validation to
assess generalization ability and mitigate overfitting.
6. Model Evaluation: Evaluate model performance using appropriate metrics
such as accuracy, precision, recall, F1-score, and area under the receiver
operating characteristic curve (AUC-ROC). Assess model robustness through
sensitivity analysis and uncertainty estimation.
7. Validation and Interpretation: Validate the trained model using the
validation dataset to ensure its generalizability to unseen data. Interpret the
model's predictions and feature importance to gain insights into the underlying
factors contributing to ASD diagnosis.
8. Model Deployment and Integration: Deploy the trained model in clinical
settings or diagnostic tools for ASD screening and assessment. Integrate the
model into existing healthcare systems while ensuring compliance with
regulatory requirements and ethical guidelines.

SITCOE,Yadrav Page 7
 Autism Spectrum Disorder Detection

5.2 Architecture diagram:

SITCOE,Yadrav Page 8
 Autism Spectrum Disorder Detection

Fig 5.2.1: Architecture Diagram

In this proposed system , we will create Autism Specturum disorder detection that
predicts the patient having Autism or not . Autism simply means it is a
neurodevelopmental diseases. Occurs in any age . using machine learning
classification algorithms we can find the autism , on the basis of autism_score.

5.3 Data Flow Diagram :

5.3.1.0 Level DFD:

SITCOE,Yadrav Page 9
 Autism Spectrum Disorder Detection

Fig 5.3.1: 0 Level DFD

5.3.2 1 Level DFD:

Fig 5.3.2:1 Level DFD

5.3.3 2 Level DFD :

SITCOE,Yadrav Page 10
 Autism Spectrum Disorder Detection

5.3.3: 2 Level DFD

5.4 Use Case Diagram :

Fig 5.4.1: Use Case Diagram

5.5 Activity Diagram:

SITCOE,Yadrav Page 11
 Autism Spectrum Disorder Detection

Fig 5.5.1: Activity Diagram

CHAPTER 6

RESULTS AND DISCUSSION

SITCOE,Yadrav Page 12
 Autism Spectrum Disorder Detection

6.1 Results:

Autism spectrum disorder (ASD) detection using machine learning typically involves
analyzing various behavioral, cognitive, and physiological data to identify patterns
associated with ASD. Machine learning algorithms can be trained on datasets
containing information such as social interactions, communication skills, repetitive
behaviors, and sensory sensitivities to develop models that can predict the likelihood
of ASD in individuals. These models can aid in early detection and intervention,
improving outcomes for those with ASD.

6.2 Screenshots of Project:

6.2.1 Pie chart for the number of data for each target

SITCOE,Yadrav Page 13
 Autism Spectrum Disorder Detection

Fig 6.2.1 Pie chart for the number of data for each target
The dataset we have is highly imbalanced. If we will train our model using this data
then the model will face a hard time predicting the positive class which is our main
objective here to predict whether a person has autism or not with high accuracy.

6.2.2 Heat map for the highly correlated features

6.2.2 Heat map

From the above heat map, we can see that there are only one highly correlated
features which we will remove before training the model on this data as highly
correlated features do not help in learning useful patterns in the data.

SITCOE,Yadrav Page 14
 Autism Spectrum Disorder Detection

6.2.3 Validation And Accuracy

Fig 6.2.3 Validation and Accuracy

Now let’s train some state-of-the-art machine learning models and compare them
which fit better with our data.

6.2.4 Prediction

Model Evaluation
From the above accuracies, we can say that Logistic Regression and SVC ()
classifier perform better on the validation data with less difference between the
validation and training data. Let’s plot the confusion matrix as well for the
validation data using the Logistic Regression model.

SITCOE,Yadrav Page 15
 Autism Spectrum Disorder Detection

6.2.4 Confusion matrix for the validation data

SITCOE,Yadrav Page 16
 Autism Spectrum Disorder Detection

CHAPTER 7

FUTURE WORK

Although autistic caregivers as well as physicians and researchers would like to know
the exact cause of ASD and find a definitive cure for it, this goal still seems to be
distant. Accordingly, the realistic achievable goal in dealing with ASD is to try to
direct all the available resources to help its sufferers to improve their skills and
functioning and get the utmost benefits of their strengths aiming at improving their
quality of life. Socioeconomic support for autistic caregivers is also essential to
empower them in helping their children across their journey for a better tomorrow.
Lastly, we must believe that “whenever and wherever there is help, there is hope”.

SITCOE,Yadrav Page 17
 Autism Spectrum Disorder Detection

CHAPTER 8

CONCLUSION

Machine learning holds great potential in revolutionizing the early detection and
diagnosis of autism spectrum disorder. Continued research, collaboration, and
innovation are essential for advancing machine learning technologies in ASD
detection. By leveraging the power of machine learning, we can improve outcomes
and quality of life for individuals on the autism spectrum.

The machine learning model we have created is around 80% to 85% accurate. The
disease for which there are no diagnostics methods machine learning models are
able to predict whether the person has Autism or not. This is the power of machine
learning by using which many of the real-world problems are being solved.

SITCOE,Yadrav Page 18
 Autism Spectrum Disorder Detection

REFERENCES

[1] S. Sugavanam, B. V. Sankar, V. K. Varadan and V. V. Varadan, “Active

Control of a Lightly Damped Cantilever Beam Using Spatial Distributed
Actuators and Sensors”, Journal Wave-Material Interaction , Vol 11, No. 1,
1996, pp. 67-84.

[2] Marek Pietrzakowski, “Experiment on a Cantilever Beam Control and

Theoretical Approximation”, Journal of Thermal and Applied Mechanics,
2002, pp. 667-686.

[3] Gustavo Luiz C. M. de Abreu, José F. Ribeiro, “Spatial H∞Control of A

Flexible Beam Containing Piezoelectric Sensors and Actuators”, ABCM
Symposium Series in Mechatronics, Vol. 1, 2004, pp. 243-253.

[4] Jin-Chein Lin, M.H. Nien, “Adaptive control of a composite cantilever beam
with piezoelectric damping-modal actuators/sensors”, Composite Structures
70, 2005, pp. 170–176.

[5] C. M. A. Vasques, J. Dias Rodrigues, “Active Vibration Control of Smart

Piezoelectric Beams: Comparison of Classical and Optimal Feedback
Control Strategies”, Computers and Structures 84, 2006, pp. 1402-1414.

[6] Zhi-chengQiu, Xian-min Zhang, Hong-xin Wu, Hong-hua Zhang, “Optimal

Placement and Active Control for Piezoelectric Smart Flexible Cantilever
Plate”, Journal of Sound and Vibration 301, 2007, pp. 521-543.

[7] Zeki KIRAL, Levent MALGACA, Murat AKDAĞ, “Active Control of

Residual Vibrations of a Cantilever Smart Beam”, Turkish J. Eng. Env. Sci.
32, pp. 51-57.

[8] Shawn Le, “Active Vibration Control of A Flexible Beam”, SJSU

ScholarWorks, 2009.

[9] Wu Dafanga, Huang Liang, Pan Bing, Wang Yuewu and Wu Shuang,
“Experimentalstudyandnumericalsimulationofactivevibrationcontrolofahigh

SITCOE,Yadrav Page 19
 Autism Spectrum Disorder Detection

lyflexiblebeamusingpiezoelectricintelligentmaterial”, Aerospace Science

and Technology 37, 2014, pp. 10–19.

[10] Soumen Das, “Vibration and surface finish in gear hobbing―some

experiments and analyses”,G.C.Sen Memorial Machine Tools and Research
laboratory MEJU, 2010.

[11] Jon Robert Pratt, “Vibration Control for Chatter Suppression with
Application to Boring Bars”, Virginia Polytechnic Institute and State
University, 1997

SITCOE,Yadrav Page 20