Predicting Dyslexia in Arabic-Speaking Children Through Handwritten Images Using Deep Learning Methods

Dyslexia is categorized as an educational disorder that affects the ability to spell, read, and write. It is considered one of the most common learning difficulties in Saudi Arabian schools. Research has demonstrated the close connection between the reading and writing processes; moreover, deficiencies or difficulties in an individual’s reading process have been associated with difficulties in writing. In this paper, we utilized a novel method to reveal dyslexia among children in elementary school in Saudi Arabia using handwritten images of alphabet letters. We present new datasets of Arabic letters written by children aged 7-10 in early childhood schools in Jeddah, Saudi Arabia. These datasets contain 3,611 characters written by two groups of children: one comprised of children who suffered from dyslexia and one without this disorder. Furthermore, we propose a Convolutional Neural Network (CNN) model for the automatic detection of dyslexia using handwriting. The results demonstrate that our model exhibits outstanding performance, attaining accuracy levels of 95.30% and 95.97% before and after the data augmentation technique, respectively, on our dataset.