Cross-Modal Biomarkers: Combining Eye Gaze and Facial Recognition for Autism Spectrum Disorder

Abstract

Autism Spectrum Disorder (ASD) is a neurodevelopmental condition characterized by challenges in social communication and repetitive behaviors. Early diagnosis is crucial for effective intervention, yet current diagnostic methods primarily rely on behavioral assessments that are subjective and time-intensive. This paper explores the integration of cross-modal biomarkers, specifically eye gaze patterns and facial recognition technologies, to enhance the accuracy and efficiency of ASD diagnosis.

Recent advancements in eye-tracking technology offer precise metrics on gaze patterns, which deviate in ASD individuals during social interactions, such as reduced fixation on faces and atypical engagement with objects. Concurrently, facial recognition systems have shown promise in detecting subtle anomalies in facial expressions and emotional responses, which are often atypical in individuals with ASD. By combining these modalities, we hypothesize that a more robust, objective framework for early ASD detection can be established.

The proposed methodology involves the collection of eye-tracking data and the application of facial recognition algorithms to analyze a dataset of both ASD-diagnosed and neurotypical subjects. Machine learning models are employed to integrate these data streams, uncovering patterns indicative of ASD. This dual approach aims to mitigate the limitations of single-modality diagnostics, offering a more comprehensive understanding of the neurodevelopmental aspects of ASD.

Preliminary results indicate that the cross-modal approach significantly improves diagnostic precision, with increased sensitivity and specificity compared to traditional methods. The integration of eye gaze and facial recognition technologies not only enhances diagnostic capabilities but also paves the way for real-time, non-intrusive screening tools. This innovative framework holds potential for widespread clinical application, improving early diagnosis and personalized intervention strategies for individuals affected by ASD.