Efficient Deep Learning Model for Detecting Dyslexia in Children from an Online Gamified Test Dataset

Abstract

Dyslexia, a neurodevelopmental disorder, significantly impacts the learning ability of young children, often going undetected until later stages of edu-cation. Early detection of this disease is essential so that the impact can be significantly reduced by prompt intervention. But because of the reliance on labour-intensive procedures and limited scalability, traditional detection methods have not been able to provide immediate and accurate assess-ments. In response, this research introduces a standardised approach utilising an online gamified test dataset to predict the risk of dyslexia through an artificial neural network model, while incorporating advanced preprocessing techniques, effectively addressing the problems of class imbalance, outliers, and noise within the data. With an accuracy of 97%, this research demonstrates the remarkable utility of this approach and surpasses the results of the state-of-the-art methods. These findings, together with our model’s improved sensitivity and specificity, highlight the revolutionary potential of our approach. This innovation is important for the field of learning disorders and neurodevelopmental research because it has the potential to improve the educational paths of many children.