Computing Multimodal Dyadic Behaviors During Spontaneous Diagnosis Interviews Toward Automatic Categorization of Autism Spectrum Disorder

Autism spectrum disorder (ASD) is a highly-prevalent neural developmental disorder often characterized by social communicative deficits and restricted repetitive interest. The heterogeneous nature of ASD in its behavior manifestations encompasses broad syndromes such as, Classical Autism (AD), High-functioning Autism (HFA), and Asperger syndrome (AS). In this work, we compute a variety of multimodal behavior features, including body movements, acoustic characteristics, and turn-taking events dynamics, of the participant, the investigator and the interaction between the two directly from audio-video recordings by leveraging the Autism Diagnostic Observational Schedule (ADOS) as a clinically-valid behavior data elicitation technique. Several of these signal-derived behavioral measures show statistically significant differences among the three syndromes. Our analyses indicate that these features may be pointing to the underlying differences in the behavior characterizations of social functioning between AD, AS, and HFA — corroborating some of the previous literature. Further, our signal-derived behavior measures achieve competitive, sometimes exceeding, recognition accuracies in discriminating between the three syndromes of ASD when compared to investigator’s clinical-rating on participant’s social and communicative behaviors during ADOS.