Module 45: Predicting health scores from EEG data

Faculty Contact: Linda Petzold

Research Areas:

Abstract:
EEG data is obtained from fitting subjects with special equipment and monitoring them while they undergo specific activities. In this case, our goal is to predict the real-time health score that is calculated by a fitness watch, given brain activity images from the EEG. We assign images a 0-1 (black-white) scaled matrix of 68 pixels. From here, we convert the images into pure data (0-1 scale) and run models. We attempt to use computer vision to extract information, but early tests find that even human beings could not decipher a pattern here; using traditional machine learning & statistical modeling was the best route to take.

Active Quarters:

  • Spring 2018: James Bird