Atypical eye gaze has been observed as an early symptom of autism spectrum disorder (ASD) and can help with early ASD screening. However, the existing eye-tracking methods are expensive and require specialized equipment. This study aims to use computational methods based on computer vision analysis to quantify eye gazing patterns and their association with ASD development.
This prospective study included a total of 1,564 toddlers aged 16 to 38 months, of which 993 completed the study. A mobile app displayed on a smartphone or tablet was used to quantify the differences in eye gaze patterns. The primary outcome of the study was eye gazing patterns analyzed by computer vision analysis.
The findings suggested that distinctive eye-gaze patterns were common in toddlers with ASD. Eye movements were characterized by a reduced gaze to salient social moments and social stimuli during movies and TV shows. Previously unknown deficits in the coordination of gaze with speech sounds were also identified. Researchers found that the area under the receiver operating characteristic curve discriminating non-ASD and ASD patients were 0.90.
The research concluded that the mobile app reliably quantified eye gazing patterns in toddlers, resulting in an early and accurate diagnosis of autism spectrum disorder.
Ref: https://jamanetwork.com/journals/jamapediatrics/article-abstract/2779395?resultClick=1
