Novel Assessment Tool for Fetal Alcohol Spectrum Disorder (FASD) Based on Eye Movement Behaviours
Fetal alcohol spectrum disorder (FASD) is the leading cause of preventable developmental disability among Canadians, affecting at least 1% of the population. Diagnosis of FASD requires collaboration from a multidisciplinary team and can be a lengthy process frequently involving long wait times. Our objective was to conduct a validation study to confirm the ability of eye tracking to efficiently and objectively identify children with FASD. In the validation study, 32 children with an FASD diagnosis and 25 typically developing control children completed three eye-tracking tasks, measuring automatic and voluntary eye movement responses, spatial working memory and visuospatial skills. Data previously collected (105 typically developing control children and 68 children with FASD) was analyzed using the same automated program, and used as the training data for the classifier. Features extracted from the eye movement control tasks were used as input to test an extreme learning machine model’s ability to accurately classify children into their respective groups. Children with FASD exhibited significant differences in eye movement performance in each of the three eye-tracking tasks, including end point error, percentage of direction errors, percentage of trials with step saccades, and overall accuracy. Classification of the validation cohort participants (FASD and control) was fairly high, producing a sensitivity of 72%, a specificity of 88%, and an overall accuracy of 79%. Features recognized as important for distinguishing between the control and FASD groups were similar or related to features that were shown to differ significantly between groups. These results support the notion that eye-tracking provides information about differences in brain function between clinical and control groups that can be used to train computational models to classify participant groups with a fairly high degree of accuracy. Measuring eye movement behaviours can help identify brain dysfunction due to prenatal alcohol exposure, providing insight into potential functional biomarkers of FASD.
URI for this recordhttp://hdl.handle.net/1974/23778
Request an alternative formatIf you require this document in an alternate, accessible format, please contact the Queen's Adaptive Technology Centre
The following license files are associated with this item: