Final year undergraduate thesis. I wanted to personalize the online learning experience for MOOCs. The goal was to develop a system that would detect periods of low attention span when a person was watching a video and create review content/quiz questions from those timestamps.
I conceptualized, designed, implemented, and evaluated the system to detect moment-to-moment fluctuation in sustained attention using a novel psychophysical task. While previous works averaged sensor readings over several seconds to classify attentional states, we were able to classify changes in attentional state at 800ms intervals with up to 81% accuracy using Electroencephalogram (EEG) and Photoplethysmogram (PPG) devices. This is the first study to classify moment-to-moment attention fluctuation at a sub-second scale using commercial grade EEG device.
This work was extended to a publication here.
Tools: Python, MATLAB