Dyslexia is a learning disability, which is a neurological condition that causes hurdles and challenges in learning. Dyslexics typically struggle with poor reading, writing, spelling, and fluency skills but this has nothing to do with their IQ. Children who are dyslexic can enhance their skills if the problem is diagnosed early. To identify dyslexia, researchers have put forth techniques, including facial image capture and analysis, eye tracking, magnetic reasoning imaging (MRI), game-based approaches, reading and writing assessments, and electroencephalography scans (EEG). This study examines recent advances in the use of machine learning algorithms to detect dyslexia, highlighting the articles published since 2010. The inclusion criteria for articles were based on the open access availability and those which involved different machine models for classification of dyslexics and non-dyslexics. A total of 45 articles were taken for review after screening manually. The analysis of the implementations using different machine learning algorithms has been done, highlighting the data preprocessing techniques, feature selection, and extraction methods applied on EEG and MRI dataset. The findings from the study have led to the conclusion that the majority of the researchers have made use of EEG data to build classification models for dyslexia, and the most preferred models used were SVM and CNN.