Automatic eye-movement-based screening for dyslexia using transformer pretraining
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Dyslexia, a common learning disability, benefits from early diagnosis. We present two models—an LSTM and a pre-trained transformer—designed to automatically classify dyslexia based on eye movements recorded during natural reading. Both models achieve the current state-of-the-art AUC of 0.94, but do not surpass it. Nevertheless, the LSTM architecture appears preferable for two reasons: it reaches this performance level without requiring pretraining, and it remains stable with limited test data.
