# Real-world application of an eye-tracking device for autism screening and diagnosis: a short report from public demonstrations in Qatar, Dubai and the U.S

## TL;DR

An eye-tracking device was tested in public events to screen for autism, showing high accuracy and potential for real-world use.

## Contribution

Demonstrated the real-world feasibility and accuracy of an eye-tracking-based autism screening tool in diverse adult populations.

## Key findings

- Median AI scores differed significantly between typically developing, neurodivergent, and ASD groups.
- The tool achieved 100% sensitivity and 99.4% specificity at the optimal cut-off score.
- 95% of participants classified as ASD had a formal diagnosis, supporting the tool's validity.

## Abstract

Social attention abnormalities are a hallmark of Autism Spectrum Disorder (ASD), often characterized by atypical gaze patterns.

This report showed the real-world feasibility of an eye-tracking–based screening paradigm for ASD across diverse general population samples.

A total of 536 adults participated in public demonstrations. An Autism Index (AI), derived from a previously validated eye-tracking paradigm, was calculated from gaze patterns toward social and nonsocial stimuli. Participants were stratified into typically developing (TD), neurodivergent, and ASD groups based on validated AI cut-offs.

Median AI scores differed significantly across groups − 0.31 (TD), 0.53 (neurodivergent), and 0.69 (ASD)- with post-hoc tests confirming higher scores in the ASD and neurodivergent groups versus TD. Gender-based analyses showed that males had significantly higher AI scores than females (p = 0.028). Among those classified as ASD, 95% reported a formal diagnosis, supporting the validity of the tool. Receiver operating characteristic (ROC) analysis demonstrated excellent diagnostic performance, with an area under the curve (AUC) of 0.997. At the optimal cut-off score of 0.555, the tool achieved 100% sensitivity and 99.4% specificity.

This report highlights the feasibility and accuracy of the eye-tracking paradigm as a scalable and objective screening tool for ASD in general adult populations, supporting its potential for broader clinical implementation.

Not applicable.