Mental State Recognition and Recommendation of Aids to Stabilize the Mind Using Wearable EEG

Abstract

Emotions play an important role in the physical activities and mental health of the human. The ability to correctly determine and interpret the mental state of a person would offer new opportunities for medical and non-medical purposes. With the fast-evolving technology and the improvements in the field of emotion recognition, numerous studies have been carried out to overcome the challenges faced during the emotional recognition. The purpose of this project is to develop a solution to recognize the current mental state of a person by analyzing Electroencephalogram (EEG), which capable of detecting the electrical activity of the brain in real-time and determine five different emotions: happiness, sadness, calm, fear, and neutral emotions. Further, the presented web application provides the best remedy to balance an unstable mindset based on the emotional state determined. Dataset has collected from 33 participants including both males and females at the age between 20-50 years. EEG signals were acquired from each participant using EEG-headband called Muse in a quiet-controlled environment. Participants were advised to watch a five minutes video clip which consists of five videos in sequence, which allocated one minute for each class of mental state and the collected datasets were used for both train and test for different emotions after applying proper pre-processing techniques. The set of features is selected from the EEG data and applied different feature selection algorithms and mental state classification algorithms to compare their recognition accuracy and performance. From the tested multiple classification methods, the Random forest classifier achieved a maximum prediction accuracy of 87.12% and used to mental state recognition. Mood, the web-based application is developed to obtain the current mental state while prompting the best set of remedies based on user feedback collected.