Constrained Multivariate Functional Principal Components Analysis for Novel Outcomes in Eye-Tracking Experiments

Area of Science:

• Neuroscience
• Developmental Psychology
• Biomarker Research

Background:

• Autism spectrum disorder (ASD) is characterized by challenges in social communication and sensory processing.
• Visual attention is a key area of interest for ASD biomarker research, measurable through eye tracking (ET).
• Current ET analyses often simplify complex gaze data, analyzing regions of interest (ROIs) separately.

Purpose of the Study:

• To develop a novel multivariate functional outcome for analyzing joint looking times across multiple ROIs in ET experiments.
• To introduce a constrained multivariate functional principal components analysis (FPCA) to capture variations in this outcome.
• To apply these methods to understand social attention differences in children with ASD.

Main Methods:

• Proposed a novel multivariate functional outcome integrating proportion looking time across multiple ROIs.
• Developed a constrained multivariate FPCA procedure accounting for the sum-to-one constraint of proportions.
• Applied the method to eye tracking data from the Autism Biomarkers Consortium for Clinical Trials (ABC-CT) Activity Monitoring task.

Main Results:

• Identified dominant modes of variation in proportion looking times across multiple ROIs for children with ASD and typically developing (TD) peers.
• Revealed richer analyses of diagnostic group differences in social attention patterns.
• Demonstrated the utility of the novel functional outcome and FPCA method in capturing complex gaze behaviors.

Conclusions:

• The novel multivariate functional approach offers a more comprehensive analysis of visual attention in ASD.
• This method enhances the understanding of social attention deficits and variations in autism.
• The findings contribute to the development of more sensitive biomarkers for ASD using eye tracking data.