Can Physiological Sensors Predict Psychological State?
This summer, computer science concentrators Sindy Liu’18, Eseosa Asiruwa’18, Mitchel Herman’19 and Matthew Goon’18 are doing research with machine learning on outputs from various sensors. The research project is directed by Stephen Harper Kirner Chair of Computer Science Stuart Hirshfield.
Data from physiological sensors, including heart rate and electrodermal activity (EDA), functional Near Infrared Spectroscopy (fNIRS), electroencephalogram (EEG), and electrocardiogram (ECG), are primarily used to detect electrical activity in the brain or heart, or electrical conductance of the skin. Current research indicates that such signals change depending on the person’s psychological state, and can be used to predict traits such as feelings, attention and awareness.
Previous research has collected and processed physiological data from different devices in different manners. This year, student researchers are attempting to collect, devise a common format for, and run data from five different sensors through a common machine learning software package (Orange Data Mining), in order to give a more comprehensive and accurate analysis of a person’s psychological state.
“Our primary goal is to combine (inputs from) all the sensors so that it can be run through machine learning (Orange). And our secondary goal is to be able to run it with a user interface,” said Liu. So far the team is working on combining data from fNIR and EEG devices. In the long run, Liu hopes that the program will be able to know what a person is thinking by analyzing their physiological state.
“I believe that computer science is practical and that by combining it with other fields, we can do so much more than just computer science by itself,” Liu concluded.