Identification of Learning Disability using ML

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056
Volume: 09 Issue: 01 | Jan 2022 www.irjet.net p-ISSN: 2395-0072
© 2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal | Page 1252
Identification of Learning Disability using ML
Nitin Ahire1, Srushti Patkar2, Tanu Prajapati3, Chetna Patil4, Vijay Raj5
1Professor, Department of Electronics and Telecommunication Engineering, Xavier Institute of Engineering,
Mumbai, Maharashtra, India
2,3,4,5Student, Department of Electronics and Telecommunication Engineering, Xavier Institute of Engineering,
Mumbai, Maharashtra, India
---------------------------------------------------------------------***----------------------------------------------------------------------
Abstract - Learning disability is a broad term that refers
to a variety of learning issues. The children with learning
disabilities are neither slow nor intellectually retarded.
Learning disability such as Dyslexia is a neurological
condition that is characterized by imprecise comprehension
of words and poor reading skills. It affects a large number of
school-aged children, with males being more likely to be
affected, placing them at risk for poor academic
performance and low self-esteem for the rest of their lives.
Our research entails developing a machine learning model
to analyse EEG signals from people with learning difficulties
and provide results in minutes with the highest level of
accuracy. The goal is to determine which data pre-
processing approaches and machine learning algorithms
are the most effective in detecting learning disabilities. The
initial stage of the project entails creating an ML model with
the help of a dataset.
Key Words: Electroencephalography (EEG), Learning
disability, Feature extraction, Classification algorithms,
Machine Learning.
1. INTRODUCTION
For a long time, the challenges of children with learning
disabilities have been a source of concern for parents and
school authorities. Children with learning disabilities can
learn successfully and subsequently become winners in
society with the correct aid at the right time, proper
assessment, and remediation. Nowadays, machine
learning is utilised to predict future events in a wide range
of industries. Predicting learning disabilities in children
and discovering the underlying causes are two of the most
beneficial applications of machine learning. There are a
variety of reasons why learning disabilities might occur.
The neurological changes in brain shape and function are
one of the most important ones. The ability to grasp
information is affected by a learning disability. The nature
of these disability is quite difficult to comprehend.
However, significant progress has been made in mapping
some of the issues that learning disability types confront,
as well as specific brain regions and structures. It is not
considered a learning disability if a person has some sort
of visual, hearing, or motor impairment. Learning
disabilities do not include mental retardation, emotional
instability, or cultural problems. One of the emerging
methods being studied for finding unique brain activation
patterns in dyslexics is electroencephalography (EEG). An
electroencephalography (EEG) is a test that uses small
metal discs (electrodes) connected to your scalp to detect
electrical activity in your brain. Even when you’re
sleeping, your brain cells communicate via electrical
impulses. In an EEG recording, this activity appears as
wavy lines. Machine learning is being employed in a wide
range of fields to forecast future outcomes. Predicting
learning disabilities in children, recognizing the actual
disability, and how early it can be recognized is one of the
most useful areas of machine learning application.
2. LITERATURE REVIEW
2.1 Advanced Machine Learning Techniques To
Assist Dyslexic Children For Easy Readability
The focus of this paper is on an evaluation of software and
hardware options that can help dyslexic children. K Means,
K Nearest Neighbor, Adaptive clustering, LS Algorithm,
Support vector machine, and Human Markov Model are
among the machine learning techniques employed.
Machine learning algorithms are commonly used to extract
speech features and determine accuracy and performance
improvement in youngsters. As a result, the study will be
focused on youngsters aged five to seven who are
experiencing difficulty reading Hindi language words. To
help dyslexic children, machine learning techniques will be
used in the design process. Two and three letters are used
here. The system is trained using a dynamic Time
Wrapping method with Hindi words as input. Once the
words have been programmed into the system. A dyslexic
youngster will be taught to read words. If a youngster
pronounces a word incorrectly the first time, he or she will
be given another chance to read. If the same occurrence
happens three times, the system will say the word out loud,
accompanied with an image, so that the youngster will
learn the term. The word uttered by the child. The same
session will be repeated for the child for another 20
minutes with various words. Machine learning techniques
will be utilised to recognise speech in this case. If the
system is tested with the same user, the Dynamic Time
Wrapping technique yields roughly 90% to 100% accuracy,
and 30% accuracy if the system is tested with a new user.

2.2 Machine Learning based Learning Disability
Detection using LMS
This study focuses on an E-learning system built with
Moodle, an open-source Learning Management System
(LMS) that allows tutors and students to collaborate more
effectively. Using dedicated courses built on the basis of
several elements of an LD student, this method recognises
two learner profiles: students with Learning Disability
(LD) and students without Learning Disability (Non-LD).
This work also includes several steps of our informal
testing approach for capturing learning factors for
Dyslexic pupils. The first stage, data collecting, has two
approaches: the first is for a smaller age group of 8-10
years with limited parameters, while the second is for a
larger age group of 11-13 years (grades 6-8), with
additional parameters. Speech-to-Text (STT) conversion
was performed on the audio responses of the users using
Natural Language Processing (NLP). The responses were
analysed using the Python programming language.
Machine Learning (ML) is utilised to determine whether or
not the user has LD (Dyslexia in this case). The binary
classification is performed using two machine learning
techniques, Logistic Regression (LR) and Support Vector
Machine (SVM), with the dataset's two classes being LD
(1) and Non-LD (0). The findings are shown for both ways,
and a comparison of the datasets generated in the final
strategy for capturing parameters using NLP
demonstrates that the dataset obtained in the final
approach is better and more robust. When it comes to
doing detection based on the created dataset, the LR
algorithm for ML outperforms SVM.
2.3 Machine Learning and Dyslexia: Diagnostic
and Classification System (DCS) for Kids with
Learning Disabilities
They proposed an automated diagnostic and
categorization system in this study. Pre-classified data
from 857 schoolchildren's spelling and reading scores
were used to train the system. To classify the data, the
twenty-fifth percentile was applied to the scores. Children
with results below the twenty-fifth percentile were
determined to have dyslexia, while those with scores
above the twenty-fifth percentile were declared non-
dyslexic. The system is made up of three parts: the
diagnostic module is a pre-screening tool that
professionals, trained users, and parents can use to
discover dyslexia signs. The second module, classification,
divides the students into two groups: non-dyslexics and
dyslexics with suspicion of dyslexia in spelling and
reading. The third module is a research analysis tool. With
98 percent accuracy, the results suggest that 23 percent of
children were at risk for dyslexia in the training data and
20.7 percent in the testing data.
2.4 Diagnosis of Dyslexia using computation
analysis
This study discusses Dyslexia diagnosis using a computer
system, taking into account people's issues with reading,
spelling, writing, and speaking. As a result, employing
dyslexia metrics methodologies, a computational analysis
classifier will be created. As a result, the Gibson test of
brain skills will be used, taking into account the effects of
working memory, auditory (hearing and speech) and
visual memory and cognition, visual and auditory
perceptions, writing and motor skills, math and time
management, behaviour, health, development, and
personality, and cognitive ability in people with learning
disabilities, particularly reading difficulties. The suggested
dataset, which contains 80 records of children, will be
analysed using computational analysis with classifiers.
This computing model was created and applied to aid in
the discovery of underlying issues that may impair
learning to read and write, as well as concerns that may
cause difficulty with memorised understanding. This
concept is used to assist counsellors and parents in
comprehending the challenge and guiding children down
the correct path to educational success.
3. PROPOSED SYSTEM
Identification of learning disability using machine learning
techniques uses four steps:
3.1 Data collection
The initial step in the detecting process is to perform a user
study and collect data from users. Behavioural elements of
participants during standardised tests, such as reading and
writing, phonological awareness, and working memory, are
examined by psychologists in traditional dyslexia diagnosis
procedures. Dyslexic people are identified by their low
scores on these exams. However, because symptoms differ
between people, these procedures are generally time-
consuming and useless for a wide group of people. As a
result, researchers adopt machine learning methodologies,
which are less time intensive and often inexpensive. EEG
headsets use electrodes placed in an array along the user's
or research subject's scalp to monitor brain activity.
3.2 Preprocessing, feature extraction and feature
selection
Before using machine learning techniques, the obtained
data must be pre-processed and filtered. This necessitates
the data being converted into either a quantitative
(numbers) or qualitative (textual categories) format. Pre-
processing is used to find relevant properties and eliminate
null values. Following the pre-processing, the feature
extraction procedure is need to be performed, in which
relevant features are discovered and a range of values is

allocated. The next step is to determine which set of
dominant qualities are most significant in defining the
object's class.
Fig -1: Proposed system
3.3 System training and classification
After feature selection, machine learning algorithms are
used to train and classify the system. The dataset is divided
into two parts: training and testing. The class information,
such as dyslexic or non-dyslexic, is contained in the
training dataset. The K-Nearest Neighbor (K-NN) and
support vector machine (SVM) will be used.
3.4 Performance evaluation
Python based tools will be used for performance
evaluation. In this case, accuracy is used for evaluating the
performance of dyslexia detection techniques using
machine learning approaches. Accuracy measures the
number of correctly classified objects to the total number
of objects.
4. CONCLUSION
The basic idea of our project is to increase the accuracy of
the learning disability assessment and reduce the time
used for learning disability assessment. Machine learning
is used in wide variety of fields and applications where
certain outcomes have to be predicted. It allows the
software applications to learn and become more accurate
in predicting outcomes. Furthermore, we will try to find
which Machine Learning algorithm works the best for
identifying specific disorder.
REFERENCES
[1] G Chakraborty, Vani. (2020). A survey paper on
learning disability prediction using machine learning.
[2] M. Modak, O. Warade, G. Saiprasad and S. Shekhar,
”Machine Learning based Learning Disability
Detection using LMS,” 2020 IEEE 5th International
Conference on Computing Communication and
Automation (ICCCA), 2020, pp. 414-419, doi:
10.1109/ICCCA49541.2020.9250761.
[3] Rehman Ullah Khan,Julia Lee Ai
Chang,Oon Yin Bee, Machine learning and
Dyslexia:Diagnostic and classification system (DCS)
for kids with learning disabilities, International
Journal of Engineering and Technology3.18(2018) 97-
10
[4] H. M. Al-Barhamtoshy and D. M. Motaweh, ”Diagnosis
of Dyslexia using computation analysis,” 2017
International Conference on Informatics, Health and
Technology (ICIHT), 2017, pp. 1-7, doi:
10.1109/ICIHT.2017.7899141.
[5] Lateef, Usman & Muniyandi, Ravie & Omar,
Khairuddin & Mohamad, Mazlyfarina. (2021). Advance
Machine Learning Methods for Dyslexia Biomarker
Detection: A Review of Implementation Details and
Challenges. IEEE Access. PP. 1-1.
10.1109/ACCESS.2021.3062709.
[6] H. Perera, M. F. Shiratuddin, K. W. Wong and K.
Fullarton, ”EEG Signal Analysis of Real-Word Reading
and Nonsense-Word Reading between Adults with
Dyslexia and without Dyslexia,” 2017 IEEE 30th
International Symposium on Computer-Based Medical
Systems (CBMS), 2017, pp. 73-78, doi:
10.1109/CBMS.2017.108.
[7] Hayes AM, Dombrowski E, Shefcyk A, et al. Learning
Disabilities Screening and Evaluation Guide for Low-
and Middle-Income Countries [Internet]. Research
Triangle Park (NC): RTI Press; 2018 Apr. Available
from:
https://www.ncbi.nlm.nih.gov/books/NBK545498/
doi: 10.3768/rtipress.2018.op.0052.1804
[8] David, Julie and Balakrishnan, Kannan. (2010).
Machine Learning Approach for Prediction of Learning
Disabilities in School-Age Children. International
Journal of Computer Applications. 9. 10.5120/1432-
1931.
[9] https://www.medicalnewstoday.com/articles/18678
7.

Identification of Learning Disability using ML

More Related Content

Similar to Identification of Learning Disability using ML (20)

More from IRJET Journal (20)

Recently uploaded (20)

Identification of Learning Disability using ML