Determining reading disorder with eye tracking and machine learning: A review of the literature

Abstract

Reading disorder, namely dyslexia, is described as the difficulty in the pronunciation and comprehension dimensions of reading. Studies in which dyslexia, one of the most common learning disorders, are examined using a technology-based and innovative technique, eye tracking, are frequently encountered. By means of eye tracking, the saccade and the fixation of dyslexic readers are reached during reading and analysis are performed with the obtained physiological data. Thus, the analysis and examination of the reading skills of individuals with dyslexia and their reading performance and profiles are revealed. In addition, in recent years, eye tracking and machine learning have started to be applied together in determining whether a reader is dyslexic or not. This review aimed to analyze and summarize the re-searches carried out to identify dyslexic individuals using eye tracking and machine learning. For this reason, in the article, after the definitions of eye movements and machine learning algorithms, studies on the detection of dyslexia in readers in four different languages, namely Spanish, Swedish, Greek and Finnish, were summarized. Therefore, it is critical to evaluate individuals with dyslexia clinically and educationally with physiological data, to diagnose them in the earliest period, to apply specific intervention programs, and to prevent academic failure and negative experiences. Thus, the accurate diagnosis can be made with-out loss of time and economic loss as a result of the application of eye tracking and machine learning even if it is complementary by clinical psychologists, guidance, psychological counseling and special education specialists in psychiatry clinics and guidance research centers. In addition to studies conducted in four different languages regarding the diagnosis of reading disorders with high accuracy using eye tracking and machine learning, individuals with dyslexia whose mother tongue is Turkish can also be evaluated and diagnosed in this way at the earliest age and specific intervention programs can be designed for them.

Keywords

• reading disorder
• dyslex-ia
• eye tracking
• eye movements
• machine learning
• diagnosis