Large Scale Eye Tracking in EU-AIMS Longitudinal Autism Research Project (LEAP): Methods and Preliminary Findings from a Biological Motion Task

Background: There are no effective treatments for the core symptoms of ASD because the underlying pathophysiology(ies) remain(s) poorly understood. Clinical trials are hampered by the profound clinical and etiological heterogeneity among individuals with ASD, as any treatment is likely only effective in some biological sub-groups. Large-scale multi-disciplinary studies and new biomarker stratification approaches are needed to identify more biologically homogeneous ASD subgroups. The EU-AIMS Longitudinal European Autism Project (LEAP) is conducted in seven European centres and will include approximately 400 individuals with ASD from 6-30 years, and 250 controls with typical development or diverse intellectual disabilities.

We present here preliminary data from a subsample (N = 197) of the LEAP dataset, focusing on one of our eye tracking tasks which measures differences in preference for biological and non-biological motion. From birth both humans and animals exhibit such a preference, suggesting that attention to biological stimuli is important for the development of effective social interaction and nonverbal communication.

Objectives: To assess preference for biological motion across adolescence and adulthood in a large international sample. We use this data to illustrate some of the processes we developed to manage the acquisition, quality control and analysis of large volumes of multi-site eye tracking data, and to discuss some of the challenges and promises of a big data approach to clinical and developmental research. We also touch on the potential for concurrent multimodal approaches to clinical and developmental research by combining behavioural, video, eye tracking, EEG and physiological (GSR, ECG) data types.

Methods: As part of a 40-minute battery of eye tracking tasks, adults (38 ASD, 51 TD) and adolescents (54 ASD, 34 TD) viewed point light displays of biological motion, paired on a trial-by-trial basis with either scrambled or rotating non-biological motion. Gaze behaviour was recorded by eye trackers and was stored, subjected to quality control, and analysed using custom-written Matlab scripts. Datasets were batch processed without manual intervention.

Results: Compared to neurotypical controls, individuals with ASD looked less at the biological motion than the control stimuli (F=6.931, p=.009, d=.309). The distribution of scores in the ASD group compared to the neurotypical group mean showed that 26.5% performed below 1SD (TD = 16.7%), and 2% performed below 2SD (TD = 1%). All participants looked longer at biological motion when it was paired with a rotating than a scrambled control stimulus (F=108.895, p<.001, d=.513) but this did not differ by diagnosis.

Conclusions: The preliminary results from the biological motion task suggest it has the potential to function as a stable biomarker in individuals with an ASD across the ages of 13-30 years. Further analyses once data collection is complete will allow us to extend this to children aged 6-12 years, and to relate these findings to ASD symptomatology and quality of life measures, in order to understand whether individuals can be stratified on this measure. The data processing pipelines developed for LEAP enabled large quantities of data to be stored and analysed efficiently in a structured and highly accessible manner.