Date of Award

31-8-2024

Document Type

Thesis

School

School of Electrical & Electroncis Engineering

Programme

Ph.D.-Doctoral of Philosophy

First Advisor

Dr.S.Jayalalitha

Second Advisor

Dr.K.Adalarasu

Keywords

Stress, Electroencephalography, Discrete Wavelet Transform, Support Vector Machine, Convolution Neural Network, Bidirectional Long Short-Term Memory

Abstract

Adolescence is a crucial part in life, and the presence of stress, anxiety, depression, and health issues during this stage is a great concern. This research aims to analyze and predict the cognitive stress in students during the examination period using EEG biomarkers. In this study, raw EEG data is acquired under two different experimental conditions, before and after examination, from 14 subjects with an eight-channel Enobio device. After preprocessing of the EEG signal, the brain rhythms such as theta, alpha, and beta sub-band energies and EEG band ratios such as neural activity, heart rate, arousal index, vigilance index and cognitive performance attentional resource index (CPARI) extracted for before and after examination conditions using db4 as mother wavelet with six level decomposition. The extracted features are analyzed before and after exams, as well as gender-wise categories, using SPSS software.

Machine learning techniques are introduced to classify two states as stress and non-stress states. The raw EEG signals are acquired from 25 subjects under two conditions, each with a 3-minute duration: non-stress (relax mode) and stress (mental task), using an EEG device. A total of eleven EEG features were extracted using the discrete wavelet transform technique. The evaluation metrics are compared for different classifier algorithms. The validation of proposed model is carried out using benchmark data from the physionet database.

Further, to improve the performance metrics of classifier algorithms, the proposed model includes hybrid features (both time-frequency and time domain features). EEG data are collected from 110 students under two conditions: relaxing and performing the mental task, each lasting 5 minutes. After pre-processing, the wavelet-based time-frequency features, such as three relative sub-band energies and eight EEG band ratios, are computed. Also, five-time domain-based statistical features such as mean, root mean square, standard deviation, skewness, and peak-to-peak value are considered.

A comparison study is carried out to analyze the performance of the classification model for wavelet-based features alone (time-frequency) and hybrid features (both time-frequency and time-domain features). The result shows that better performance metrics are obtained for the cubic SVM classifier using hybrid features. The validation of the proposed model is carried out with a benchmark database from the physionet, which consists of 36 subjects’ EEG data under relax and task state. Also, we study the importance of fixed and sliding window concepts; the result reports better performance metrics for the sliding window approach using hybrid features.

A deep learning technique is introduced for the classification of stress and non-stress (relax) states using hybrid features with a 2-second sliding window approach. The CNN-BLSTM model reported the highest accuracy for the classification of stress and non-stress states. Finally, an experimental study for the multiclass classification of stress levels such as relaxed, low, medium, and high-stress is carried out by acquiring EEG signals from 35 subjects by performing different levels of task, each with a 5-minute duration. A sliding window approach of 2 seconds with 50% overlap is used for feature extraction. Hybrid features are considered and fed to CNN and BLSTM for classification, and better performance metrics are obtained for the proposed model. In a comparison of both machine learning and deep learning approaches for the classification of stress and non-stress states, better performance metrics are reported for the deep learning method. This proposed model could be used to support the health care providers in early diagnosis of stress among students to avoid suicidal thoughts and provide necessary treatment or counselling in advance.

Download

Share

COinS