Title: Pilot Study to Discover Candidate Biomarkers for Autism based on Perception and Production of Facial Expressions

Abstract: Purpose: Facial expression production and perception in autism spectrum disorder (ASD) suggest potential presence of behavioral biomarkers that may stratify individuals on the spectrum into prognostic or treatment subgroups. Construct validity and group discriminability have been recommended as criteria for identification of candidate stratification biomarkers.
Methods: In an online pilot study of 11 children and young adults diagnosed with ASD and 11 age- and gender-matched neurotypical (NT) individuals, participants recognize and mimic static and dynamic facial expressions of 3D avatars. Webcam-based eye-tracking (ET) and facial video tracking (VT), including activation and asymmetry of action units (AUs) from the Facial Action Coding System (FACS) are collected. We assess validity of constructs for each dependent variable (DV) based on the expected response in the NT group. Then, the Boruta statistical method identifies DVs that are significant to group discriminability (ASD or NT).
Results: We identify one candidate ET biomarker (percentage gaze duration to the face while mimicking static 'disgust' expression) and 14 additional DVs of interest for future study, including 4 ET DVs, 5 DVs related to VT AU activation, and 4 DVs related to AU asymmetry in VT. Based on a power analysis, we provide sample size recommendations for future studies.
Conclusion: This pilot study provides a framework for ASD stratification biomarker discovery based on perception and production of facial expressions.