An Automated Framework of Stress Detection based on EEG Signals

Authors

  • Prashant Lahane, Sushila Palwe

DOI:

https://doi.org/10.17762/msea.v71i4.1327

Abstract

Stress is one type of universal emotions faced by everyone. Several factors are responsible for the stress which impact on the low performance of the individual. It also effects on the psychological and physical well being of the person. The more extended period of stress gives depression and suicidal risk.  The traditional methods of stress detection were used signals of speech,  physiological and facial expression. The traditional methods are less accurate and give problems due to exterior influences such as room temperature, sweating, limb movement, anxiety. A method is required which should be non-invasive, precise, accurate and reliable. EEG signals are the most suitable for stress detection due to its strong correlation with the stress. This paper aims to detect real-time stress based on emotion detection.  The EEG signal is acquired using a Neurosky mind wave device, and relevant features are extracted from the time space to the recurrence area using an Alpha Beta frequency Cepstral Coefficient (ABFCC). The separated highlights are characterized into glad as well as angry emotions using classifiers like K-Nearest Neighbour (KNN), Support Vector Machine (SVM), Naive Bayes (NB), decision tree and neural network (NN). Happy and irate feelings are liable for the focused and unstressed condition of the person. As glad emotions are well thought-out a relaxed, and angry emotions are measured as a focused on state.  The results show that real-time stress detection using ABFCC and KNN gives the accuracy of 90% for alpha and 92.3% for a beta band which comparatively better than the existing KDE and RER.

Downloads

Published

2022-12-27

How to Cite

Prashant Lahane, Sushila Palwe. (2022). An Automated Framework of Stress Detection based on EEG Signals . Mathematical Statistician and Engineering Applications, 71(4), 7101–7112. https://doi.org/10.17762/msea.v71i4.1327

Issue

Section

Articles