Graph Neural Networks and sequential architectures for autism detection from eye-tracking biomarkers: a multi-site study
Nisrine El Ayat, Mohammed Boutalline, Adil Tannouche · 2026
Read Graph Neural Networks and sequential architectures for autism detection from eye-tracking biomarkers: a multi-site study by Nisrine El Ayat, Mohammed Boutalline, Adil Tannouche (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13471379.txt.
Description
El Ayat and colleagues tested whether eye-tracking patterns, measured while people view images, can help detect autism spectrum disorder with artificial intelligence. Their 2026 multi-site study combined two European datasets, 57 participants from France and 27 from Italy, giving 84 people in total, 40 of them autistic.
Four model types were compared, and a memory-based LSTM network scored highest, with an AUC of 0.859 on the ranking scale where 1.0 is perfect. A graph neural network with attention nearly matched it at 0.855 and reached the best specificity, correctly clearing 93.2% of non-autistic participants. Attention mechanisms, often assumed helpful, did not significantly lift the graph model's results. The authors argue either design could support objective screening where specialists are scarce.
Catalog record
| CREATOR | Nisrine El Ayat, Mohammed Boutalline, Adil Tannouche |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 91 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |