Automatic detection of dyslexia based on eye movements during reading in Russian

Dyslexia, a common learning disability, requires an early diagnosis. However, current screening tests are very time-and resourceconsuming. We present an LSTM that aims to automatically classify dyslexia based on eye movements recorded during natural reading combined with basic demographic information and linguistic features. The proposed model reaches an AUC of 0.93 and outperforms the state-of-the-art model by 7 %. We report several ablation studies demonstrating that the fixation features matter the most for classification.