The hidden patterns in visual behavior
The hidden patterns in visual behavior
The Hidden Patterns in Visual Behavior explores how artificial intelligence can advance translational vision science by helping us analyse, restore, and understand human vision in real-world and clinical contexts.
Thesis of Ali Rahmani Nejad develops computational models for three connected domains: eye-movement analysis, prosthetic vision, and visual field loss. First, it introduces methods for automatically classifying gaze events during natural behaviour, where people move freely and use head-mounted eye-trackers. These methods address the limitations of traditional eye-movement algorithms that were mainly designed for controlled laboratory settings.
Second, the thesis investigates AI-driven approaches for simulated prosthetic vision, showing how gaze-centred encoding can improve object recognition for future visual implants.
Third, it applies machine-learning tools to identify meaningful patterns of binocular visual field loss in glaucoma and relate them to patients’ quality of life. Together, the work shows how AI can uncover hidden structure in visual behaviour and support more natural experiments, better assistive technologies, and more personalised clinical care.