A study on dyslexia detection using machine learning techniques for checklist, questionnaire and online game based datasets

Abstract

Learning disabilities are one of the most common developmental disorders in children. Learning is fundamental to a child's overall development. Children struggle with daily activities such as reading, speaking, organizing things, and so on. The specific learning disorders are classified into dyslexia, dysgraphia, and dyscalculia. Children who find difficulty in reading and are unable to differentiate speech sounds are said to have dyslexia. Dysgraphia and dyscalculia deal with written and mathematical calculations. Early diagnosis and detection are essential for early recovery from diseases. The proposed article presents methodologies and techniques used for detecting dyslexia. The primary contribution of this paper is a comparative analysis of various machine learning algorithms for diagnosing dyslexia, including SVM, KNN, Logistic Regression, K-mean Clustering, Oversampling, and Ensemble methods. Deep learning methods such as CNN and LeNet architecture have been used to identify dyslexia. The proposed study examines recent advances in detecting dyslexia using machine learning and deep learning approaches and identifies prospective research areas for the future.