Assessing Users’ Attention Classification Accuracy from EEG Data

Graduation Year

2024

Document Type

Thesis

Degree

M.S.C.S.

Degree Name

MS in Computer Science (M.S.C.S.)

Degree Granting Department

Computer Science and Engineering

Major Professor

Marvin Andujar, Ph.D.

Committee Member

Shaun Canavan, Ph.D.

Committee Member

Julia Woodward, Ph.D.

Keywords

Alpha Waves, Beta Waves, Mobile Application, Quantified-self, Theta Waves

Abstract

Determining whether a person is attentive while performing a task is a key factor for understanding why they are achieving specific results. In addition, tracking these results is an important aspect to improving them. Quantified self is a field that consists of self-monitoring to gain a better understanding of your habits, health, and overall well-being. Examples of this method include tracking sleep patterns, daily steps, mood swings, and food intake.

This thesis aims to explore the potential of using an innovative self-quantification method that utilizes a mobile application that leverages alpha, beta, and theta waves to assess college students' attention levels. Brain waves, which represent different cognitive states, have distinct frequencies and functions and can be measured through a non-invasive technique using the Enobio electroencephalogram (EEG) headset. Applying such a technique through a user-friendly mobile app holds the potential to revolutionize the way we assess attention classification and tracking. It aims to assist students to assess and track their levels of attention.

Data was collected from participants at the University of South Florida for two weeks. Participants initially performed pre-determined tasks and reported their own attention levels with a rating system and survey. The participant’s self-assessment ratings were then compared to the score generated by the attention score algorithm to display the information as a score-based system that ranges from 1 to 100. A score of 1 means the lowest level of attention and 100 means fully attentive in the task at hand, to be easily understandable. The model achieved an accuracy of 73.34% as user’s measured attention levels closely matched how they classified themselves. Therefore, such results show the benefits of using the application to assess and track the individual’s attention while performing a task.

This document is currently not available here.

Share

COinS