Eye-Tracking Biomarkers for Autism

Autism Spectrum Disorder (ASD) is marked by significant heterogeneity among individuals, presenting a major obstacle to the development and evaluation of new interventions. Progress in clinical trials for ASD has been severely limited by a lack of objective and reliable measures—known as biomarkers—that can quantify intrinsic features of the disorder or reliably track response to treatment.

The Autism Biomarkers Consortium for Clinical Trials (ABC-CT) study conducted by Frederick Shic and colleagues is one of the most rigorous evaluations of candidate eye-tracking biomarkers for ASD. The primary research question asked whether a battery of eye-tracking (ET) tasks—administered using high-precision EyeLink eye-tracking technology—could yield feasible, reliable, and clinically meaningful biomarkers for autism, suitable for use in clinical trials. The study investigated the feasability of recording eye movements, and whether ET measures demonstrated construct validity, short-term stability, group discrimination between ASD and typically developing (TD) children, and clinically relevant associations with behavioral characteristics.

Research Question and Methodology with High-Precision Eye Tracking

The multisite study included 280 children with ASD and 119 TD children aged 6–11 years. All eye-tracking data were collected with SR Research EyeLink 1000 Plus remote systems running at 500 Hz, ensuring high temporal resolution and spatial accuracy. Stimuli were presented on standardized displays, and data collection followed strict protocol manualization and centralized quality control.

The ET battery consisted of five viewing tasks:

1. Activity Monitoring: Videos of two actors interacting in a shared activity
2. Social Interactive: Videos of two children in interacting or parallel play
3. Static Social Scenes: Static photos depicting everyday scenes of social interactions
4. Biological Motion Preference: Point-light videos depicting human actions paired with non-human actions
5. Pupillary Light Reflex (PLR): Small central animation on a dark screen followed by a flash of white

The first three tasks focused on social attention and were combined into the study’s primary biomarker outcome—the Oculomotor Index of Gaze to Human Faces (OMI), which measured the proportion of gaze directed to human faces across dynamic and static social scenes.

Key Findings and Biomarker Viability

• Exceptional Data Acquisition Feasibility: Data acquisition feasibility refers to how successfully and consistently a study can collect high-quality data from participants. Across all tasks, data acquisition and validity exceeded 95% for both ASD and TD groups—an essential requirement for potential biomarker use in clinical trials. This performance highlights both the study’s rigorous protocols and the reliability of the EyeLink system in capturing high-quality ET data even in heterogeneous clinical populations.
• Strong Construct Validity: TD children consistently looked at human faces at rates far above chance in all social-attention tasks, confirming that these paradigms elicited expected visual behaviors. The PLR task also reliably produced expected pupil constriction responses. Although construct validity was evaluated primarily in the TD group, the ASD group showed comparable patterns, indicating the measures are also applicable to ASD.
• High Six-Week Stability: High six-week stability refers to how consistently a biomarker or measurement produces similar results when the same individuals are tested again six weeks later, assuming their underlying abilities or symptoms have not meaningfully changed. More formally, a measure shows a high intraclass correlation coefficient (ICC ≥ .75) between Time 1 and Time 2 assessments conducted six weeks apart. OMI and related gaze-to-face measures showed high stability over six weeks (ICC ≈ .83), supporting their potential as stable trait-level markers suitable for short-term clinical trials. PLR measures also demonstrated strong stability.
• Robust Group Discrimination: Children with ASD showed significantly reduced gaze to faces across all social tasks, with medium to large effect sizes, replicating long-standing findings in the ASD literature. These group differences remained significant even after controlling for age, IQ, site, and data quality.
• Clinically Meaningful Associations: Lower OMI scores were associated with higher autism symptom severity, lower verbal IQ, reduced communication skills, and poorer memory for faces, suggesting that the biomarker meaningfully indexes individual variability in social-communicative functioning.

The Role of EyeLink Eye Tracking Technology in Autism Research

The ABCT-CT included five testing sites that used EyeLink 1000 Plus eye tracker systems because of the high sampling rate (up to 2000 Hz binocularly), spatial accuracy, and remote mode function that enables participants to move their heads freely while testing. The results is the reliable collection of data from school-aged children, including those with attentional or regulatory challenges. These advantages have made EyeLink systems a go-to choice for autism and developmental researchers worldwide.

This study demonstrates that eye-tracking—particularly gaze-to-face measures captured with EyeLink systems—meets key psychometric requirements for potential biomarker deployment in ASD clinical trials. The strong performance of the OMI has already led to its acceptance into the FDA Biomarker Qualification Program, marking a significant milestone in the translation of eye-tracking science into clinical application.

有关眼动追踪如何帮助您的研究的信息，请查看我们的解决方案和产品页面或联系我们。我们很乐意为您提供帮助！