Computer-Aided Diagnosis of Dyslexia Using Machine Learning

Dyslexia is identified as a typical learning disorder which affects children due to which they develop difficulty in pronouncing words, writing with the correct spellings, etc. The numerical dataset includes results from a group of dyslexic people and a control group of people who can read normally. The objective here is to test how dyslexia can be diagnosed using machine learning effectively. We build models that can precisely categorize people as dyslexic or not using a variety of ML algorithms, such as random forest, LR, kernel SVM, multilayer perceptron, etc. We assess each model's performance using metrics including recall, accuracy, and precision. Based on test results, our findings show that ML algorithms are capable of accurately identifying people who have dyslexia. The model with the best results had an accuracy of 90% achieved by Naïve Bayes and logistic regression. These results imply that ML-based techniques have the potential to support early dyslexia detection, resulting in more prompt interventions and better outcomes for dyslexics.