A Convolutional Neural Network Model to Detect Stress Levels in Electroencephalogram Signals

Abstract

Chronic stress is recognized as a major contributing factor to the impairment of people's health and quality of life due to the negative impacts it causes both in personal and professional life. There is a tendency for the increasing of concerns about mental health and investments in healthcare in the coming years and this fact has motivated the conduction of several studies that aim to develop technological solutions that can be applied for monitoring of people's mental state to early detection and treatment of stress. Machine Learning and Deep Learning techniques have proven to be a promising alternative to provide technology to help solve the challenges in the medical field. This study presents a Convolutional Neural Network (CNN) model for classifying stress levels in data from electroencephalogram signals. The model was trained with a dataset generated from an experiment in which individuals were monitored by electroencephalogram while performing stress-inducing tasks. The model achieved an accuracy rate of 98.95% in predictions and presented an adequate balance between performance and computational costs. The results obtained demonstrate the potential of Deep Learning techniques as tools to aid in the health monitoring and diagnosis. Key Words: machine learning, deep learning, mental stress, electroencephalogram