More than 40 million adults in the United States are estimated to have dyslexia, but only about 2 million have ever been formally diagnosed. Existing diagnostic tools are expensive, slow, and require trained specialists, which means most adults who struggle with reading never find out why. This project investigates whether smartphone cameras can be used to screen for dyslexia in adults through eye tracking. Eye-tracking research has shown that dyslexic readers display measurable differences in reading behavior, including longer fixation durations, more frequent regressions, and irregular eye-movement patterns. Prior work in Dr. Fletcher’s group has shown that a front-facing camera on a consumer tablet, paired with a convolutional neural network, can do reliable gaze estimation without specialized hardware. I will extend that approach to adult dyslexia screening, design a model architecture that can run on-device in real time, and build a working mobile app that delivers an assessment in under ten minutes on common phones. The project will also include interviews with Boston-area teachers and adults who have been diagnosed or suspect they have dyslexia, which will inform both the app design and data collection protocol.

I am participating in SuperUROP because I want to apply my background in computer science and AI to a longer-term research project in health. I have built predictive ML models during my internship at TASK, designed a Python and data visualization curriculum for high school students through my Media Lab UROP, and developed a computer vision research project this semester in 9.60 with Prof. Sinha. My passion lies in applying AI to impactful fields, especially health and well-being, and this project would let me work on something that is grounded in a problem that I can see the high impact it can have for individuals worldwide.