Automated Quantification of Affect Synchrony: Links to Autism Risk and Social Communication in Infancy

Interpersonal affect synchrony (IAS), the moment-to-moment coordination of emotional expressions in parent-child interactions, is a key marker of early social development. However, scalable methods for quantifying IAS and examining its link to Autism Spectrum Disorder (ASD) are limited. This study aimed to evaluate whether automated, computer-vision-based quantification of IAS could capture early dyadic interaction patterns linked to an elevated likelihood of later ASD outcomes.

We analyzed video recordings of 70 mother-infant dyads at elevated risk for ASD (aged 6-14 months). IAS was quantified using a validated computer vision pipeline benchmarked against manual coding. Clinical Best Estimate (CBE) diagnoses at 18-24 months determined ASD outcomes. Associations between IAS, maternal positive responsiveness, and concurrent social-communicative functioning (assessed via the CSBS) were examined. Classification models incorporating dyadic metrics were compared to models relying solely on infant behavior.

Infants later diagnosed with ASD exhibited significantly lower IAS, despite higher maternal positive affect. IAS independently predicted concurrent social-communicative functioning. Classification models incorporating dyadic metrics demonstrated more balanced and robust performance in identifying ASD risk compared to models based solely on infant behavior.

Automated quantification of IAS provides a reliable, scalable approach to capture early dyadic interaction patterns associated with ASD risk. These findings highlight the potential of targeting parent-infant synchrony in early screening and intervention strategies, offering a promising direction for objective, behavior-based assessments in infancy.