Developing a Dyslexia Indicator Using Eye Tracking
summary
Uses the Provo Corpus of eye movements (84 native English readers, age 18–25, 63 columns and ~230,000 rows) to extract fixation durations, saccade metrics and regressions, cleans and engineers durations, then clusters with hierarchical clustering (Euclidean distance, Ward linkage, elbow method) to reveal a severity spectrum. A Random Forest Classifier with forward selection cuts features from 44 to 12, runs 9‑fold CV and reports ~88.58% accuracy (F1 ≈ 88.4%), claiming a lightweight, noninvasive dyslexia indicator.
AI-written summary · about
chapters (8)
- 0:00Dyslexia and eye-tracking intro
- 1:46History of dyslexia diagnosis
- 3:28Dataset selection and prep
- 5:15Key features for detection
- 7:21Cluster analysis of reading
- 8:23Model training and evaluation
- 12:33Results and applications
- 13:13Q&A
related talks
45mEye tracking in bilingual neurolinguistics
37mCan you Read This? How I Improved Text Readability Online for Visually Impaired
48mEye Tracking (a Deep Learning Approach)
16mEye-Trace: Segmentation of Volumetric Microscopy Images with Eyegaze
40mThe eye of the Python, an eye tracking system. From zero to...
12mThe Potential of Cognitive Circles to Measure Mental Load
28mCan we measure depression
06mSelf-supervised Learning for Reading Activity Classification