Machine Learning Classification Of Autism Spectrum Disorder Based On Reciprocity In Naturalistic Social Interactions

Abstract

AbstractAutism spectrum disorder is characterized by impaired social communication and interaction. As a neurodevelopmental disorder typically diagnosed during childhood, diagnosis in adulthood is preceded by a resource-heavy clinical assessment period. The ongoing developments in digital phenotyping give rise to novel opportunities within the screening and diagnostic process. In this study, we investigated videos of naturalistic social interaction between autistic and non-autistic adults on their predictiveness for autistic behaviors. Non-autistic control participants were either paired with each other or an autistic participant to engage in two conversational tasks. We used existing computer vision algorithms to extract information based on the synchrony of movement and facial expression. These were subsequently used as features in a support vector machine learning model to predict interaction dyad membership. Results showed high predictive accuracy of synchrony in facial movements, underlining the distinctive nature of non-verbal behavior in autism and its feasibility for digitalized assessment.