A Machine Learning Approach for Scalable Early-Age Dyslexia Detection Correlating Phonological Speech Analysis and Eyeball Tracking
Keywords:
dyslexia analysis, eyeball tracking, fixations, saccades, phonological testingAbstract
Dyslexia concerns a persistent and unexpected difficulty in developing age and experience
appropriate word reading skills, which encompasses proficiency in accuracy and efficiency. It is termed a
reading disability that at an early age is extremely hard to trace because the degree and type of dyslexia are
unique to an individual, their genetics and environmental factors. While there is yet no full understanding of the
cause of dyslexia, or agreement on its precise definition, it is certain that many individuals suffer persistent
problems in learning to read for no apparent reason. Since this could drastically affect literacy skills and
academic engagement, there is a need to address the quantifiable screening process for dyslexia as early,
effectively, and objectively as possible.
Our solution attempts to screen dyslexia problems for children at every age. This mechanism correlates an
eyeball tracking and speech-recognition technology, written in python with Matplotlib as the main plotting
library. The solution utilised is a scalable measure of dyslexia at every age. While the user is asked to read a
passage on the screen, the camera traces the pupil’s positions with respect to eye momentum and direction,
recording the independent variables- fixations and saccades. By analyzing the distribution pattern of existing
datasets, the model makes a prediction. The user is then asked to read aloud a series of 25 words, while the
algorithm processes the speech and identifies any discernible differences in speech-features (associated with
dyslexia). The independent variables (input parameters) are reaction time, backtracking, unexpected
pronunciation and reading time. The final algorithm prediction provides results on the predicted level of
dyslexia.
This novel solution aims to give estimative predictions by combining eyeball tracking (reading) and speech
recognition (phonology) so that the early detection of dyslexia leads to early intervention
References
"Dyslexia FAQ." Yale Dyslexia. Accessed
November 13, 2020.
https://dyslexia.yale.edu/dyslexia/dyslexia-faq/.
Casalis, S. (2004). The concept of dyslexia. In
Handbook of children’s literacy (pp. 25’ure, and
prospects. In Proceedings of the world congress on
dyslexia on reading, perception, and language (Vol.
, p. 272)
Desroches, Amy S., Marc F. Joanisse, and Erin
K. Robertson. "Specific Phonological Impairments
in Dyslexia Revealed by Eyetracking." Cognition
, no. 3 (2006).
doi:10.1016/j.cognition.2005.09.001.
Jones, Manon W., Mateo Obregón, M. Louise
Kelly, and Holly P. Branigan. "Elucidating the
Component Processes Involved in Dyslexic and
Non-dyslexic Reading Fluency: An Eye-tracking
Study." Cognition 109, no. 3 (2008): 389-407.
doi:10.1016/j.cognition.2008.10.005.
Mahmoodi-Shahrebabaki, M. (2014). E-learning
in Iran as a developing country: challenges ahead
and possible solutions.International Journal of
Research in EducationMethodology,6(2), 788-795
Mahmoodi-Shahrebabaki, M. (2018, February).
Dyslexia: past, present, and future. Paperpresented
at 6th Annual MTSU Literacy Research
Conference, Murfreesboro, TN.
NilssonBenfatto M, qvistSeimyr G, Ygge J,
Pansell T, Rydberg A, Jacobson C (2016)
Screening for Dyslexia Using Eye Tracking during
Reading. PLoS ONE 11(12): e0165508
