Development and Validation of Objective, Eye Tracking-Based Risk and Symptom Measures for Autism

Background: At present, there are no validated, objective, quantitative, scalable assessment tools for autism spectrum disorder (ASD).

Objectives: To develop and validate eye tracking-based measures for estimating ASD risk and quantifying autism symptom levels.

Methods: Patients were recruited into this cross-sectional, case-control study from a tertiary care, multi-disciplinary, autism diagnostic evaluation clinic. Eye tracking data were collected during a single evaluation visit with administrators blinded to all clinical information. Participants included 201 children referred for evaluation of possible developmental disorder who completed a valid eye tracking assessment (ages 1.6-17.6; 80% male; ASD n=91, non-ASD n=110). Consensus clinical diagnoses were given by the multidisciplinary team based on Autism Diagnostic Observation Schedule-2 (ADOS-2) results, developmental history, physician evaluation, and additional clinical measures. Participants viewed a 5-minute video that included 44 dynamic stimuli from 7 distinct paradigms while gaze was recorded. Five gaze metrics (glances, fixation count, fixation duration percent, first fixation duration, and average fixation duration) were computed for temporally-defined regions-of-interest within each stimulus. Autism risk and symptom indices were created by aggregating across gaze measures showing significant bivariate relationships with ASD diagnosis and ADOS-2 symptom severity levels in a training sample (75%, n=150). Receiver operating characteristic curve analysis and non-parametric correlations were used to cross-validate identification of ASD diagnosis and autism symptom severity in a test sample (25%; n=51).

Results: The autism risk index had excellent accuracy for identifying ASD diagnosis in the training sample (AUC=.92, 95%CI=.88-.96) and maintained high accuracy in the test sub-sample (AUC=.86, 95%CIs=.75-.95; Figure 1). Autism symptom indices were strongly associated with ADOS-2 total, social affect, and restricted/repetitive behavior severity scores (smallest r=.26, p=.040). The autism risk index had high internal consistency reliability (α=.92) and wide quantitative range, with 95% of non-ASD cases falling from z=-2.3 to 1.6 and 95% of ASD cases falling from z=-0.1 to 5.0. The majority of missed cases (68%) fell within +/-0.75 SD of the optimal cut point z=0.74. Autism symptom indices had high internal consistency reliability (α>=.93). Validity of autism risk and symptom indices was not substantively attenuated after adjustment for language, non-verbal cognitive ability, or other psychopathology symptoms (r=.40-.67, p>.001), indicating that the eye-tracking based measures were highly specific to autism.

Conclusions: Eye tracking measures may be useful quantitative, objective measures of ASD risk and autism symptom levels. Future studies with large, multi-site samples are needed to replicate these findings and determine their generalizability, including resistance to sampling differences and minor procedural variations. If replicated and scaled for clinical use, eye tracking-based risk and symptom measures could be used to inform clinical judgment regarding ASD identification and to track autism symptom levels in clinical trials and longitudinal studies.