EYE-C: Eye-Contact Robust Detection and Analysis during Unconstrained Child-Therapist Interactions in the Clinical Setting of Autism Spectrum Disorders

Abstract

The high level of heterogeneity in Autism Spectrum Disorder (ASD) and the lack of systematic measurements complicate predicting outcomes of early intervention and the identification of better-tailored treatment programs. Computational phenotyping may assist therapists in monitoring child behavior through quantitative measures and personalizing the intervention based on individual characteristics; still, real-world behavioral analysis is an ongoing challenge. For this purpose, we designed EYE-C, a system based on OpenPose and Gaze360 for fine-grained analysis of eye-contact episodes in unconstrained therapist-child interactions via a single video camera. The model was validated on video data varying in resolution and setting, achieving promising performance. We further tested EYE-C on a clinical sample of 62 preschoolers with ASD for spectrum stratification based on eye-contact features and age. By unsupervised clustering, three distinct sub-groups were identified, differentiated by eye-contact dynamics and a specific clinical phenotype. Overall, this study highlights the potential of Artificial Intelligence in categorizing atypical behavior and providing translational solutions that might assist clinical practice.

C1q/TNF-related protein-1: Potential biomarker for early diagnosis of autism spectrum disorder

“…62 Despite this, ASD diagnosis is usually delayed until school age especially in mild to moderate cases of ASDs 63,64 ; since the early signs, such as repetitive movements or delayed talking might be misleading and considered normal by the parents or caregivers, or it may be misdiagnosed with attention deficit hyperactivity disorder (ADHD). [65][66][67][68] Although early diagnosis of ASD even in infancy is crucial for the proper management of this condition, there are limited to no accepted clinical biomarkers for its accurate diagnosis; instead, the current global approach being practiced for ASD diagnosis is majorly made based on interviews and standard questioners such as the DSM5 rather than laboratory tests. 10,65 Since the awareness for ASD has been raised in the world and especially in developing countries, finding a reliable diagnostic biomarker has been the focus of researchers in the field.…”

A Review of and Roadmap for Data Science and Machine Learning for the Neuropsychiatric Phenotype of Autism

“…Breaking from the more standard practice of using highly structured interactions and in-lab settings for data collection, Varma et al ( 44 ) measured eye gaze differences between cohorts using network analysis from crowdsourced data collected during use of a mobile autism therapeutic, finding a statistically significant difference between the groups for a single area of interest. Alvari et al ( 45 ) analyzed eye contact during unconstrained therapist–child interactions by applying unsupervised clustering on data from 62 children with autism, identifying three distinct subgroups defined by eye contact dynamics. In one of the largest autism-related eye tracking studies containing 563 subjects with autism and 1,300 other subjects, children with autism exhibited a higher percent fixation to dynamic geometric images compared with other children ( 46 ).…”

The assessment of children with autism based on computer vision

“…Pierce and Murias et al 14,15 studied the attention of autistic children by tracking eye movements with an eye tracker; Chang and Bovery 16,17 collected gaze and attention data using convenient devices such as cameras or smartphones, and used the computer vision algorithms to analyze the attention of the autistic children; Vabalas et al 18 studied the similarity of motion imitation between the autistic and normal adults, and classified normal adults and autistic patients by using eye gaze features. Alvari et al 19 designed a software tool "EYE-C".…”