NeuroEye: AI-Powered Eye Tracking for Mental Health Detection

Abstract

Mental health remains one of the most underaddressed areas in global healthcare. Conditions such as depression, anxiety, ADHD, and chronic stress affect a significant portion of the population, yet countless individuals go through life without any form of screening or diagnosis. A major reason for this is that existing detection methods depend almost entirely on self-reported data — an approach that frequently breaks down when stigma, denial, or limited self- awareness gets in the way.To bridge this gap, we developed NeuroEye, a web-based application that identifies early indicators of mental distress by passively tracking how a person's eyes move and blink during a session. Rather than relying on questionnaires or interviews, the system uses a standard webcam to silently observe natural eye behaviour. It applies to the MediaPipe Face Mesh library to extract facial landmarks in real time and computes the Eye Aspect Ratio (EAR) to monitor blinking activity. Gaze direction is determined through iris position tracking across nine predefined zones. These combined signals are evaluated against established clinical thresholds to flag potential signs of stress, low mood, fatigue, or attention irregularities.A key design decision was keeping all computation on the user's own device — no data is transmitted to any external server, making the system entirely private. NeuroEye is not intended to replace professional medical evaluation.