An enhanced stress indices in signal processing based on advanced matthew correlation coefficient (MCCA) and multimodal function using EEG signal
Stress is a response to various environmental, psychological, and social factors, resulting in strain and pressure on individuals. Categorizing stress levels is a common practise, often using low, medium, and high stress categories. However, the limitation of only three stress levels is a significan...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2023
|
| Subjects: | |
| Online Access: | http://umpir.ump.edu.my/id/eprint/39296/1/ir.An%20enhanced%20stress%20indices%20in%20signal%20processing%20based%20on%20advanced%20matthew%20correlation%20coefficient%20%28mcca%29%20and%20multimodal%20function%20using%20eeg%20signal.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Stress is a response to various environmental, psychological, and social factors, resulting in strain and pressure on individuals. Categorizing stress levels is a common practise, often using low, medium, and high stress categories. However, the limitation of only three stress levels is a significant drawback of the existing approach. This study aims to address this limitation and proposes an improved method for EEG feature extraction and stress level categorization. The main contribution of this work lies in the enhanced stress level categorization, which expands from three to six levels using the newly established fractional scale based on the quantities' scale influenced by MCCA and multimodal equation performance. The concept of standard deviation (STD) helps in categorizing stress levels by dividing the scale of quantities, leading to an improvement in the process. The lack of performance in the Matthew Correlation Coefficient (MCC) equation is observed in relation to accuracy values. Also, multimodal is rarely discussed in terms of parameters. Therefore, the MCCA and multimodal function provide the advantage of significantly enhancing accuracy as a part of the study's contribution. This study introduces the concept of an Advanced Matthew Correlation Coefficient (MCCA) and applies the six-sigma framework to enhance accuracy in stress level categorization. The research focuses on expanding the stress levels from three to six, utilizing a new scale of fractional stress levels influenced by MCCA and multimodal equation performance. Furthermore, the study applies signal pre-processing techniques to filter and segregate the EEG signal into Delta, Theta, Alpha, and Beta frequency bands. Subsequently, feature extraction is conducted, resulting in twenty-one statistical and non-statistical features. These features are employed in both the MCCA and multimodal function analysis. The study employs the Support Vector Machine (SVM), Random Forest (RF), and k-Nearest Neighbour (k-NN) classifiers for stress level validation. After conducting experiments and performance evaluations, RF demonstrates the highest average accuracy of 85%–10% in 10-Fold and K-Fold techniques, outperforming SVM and k-NN. In conclusion, this study presents an improved approach to stress level categorization and EEG feature extraction. The proposed Advanced Matthew Correlation Coefficient (MCCA) and six-sigma framework contribute to achieving higher accuracy, surpassing the limitations of the existing three-level categorization. The results indicate the superiority of the Random Forest classifier over SVM and k-NN. This research has implications for various applications and fields, providing a more effective equation to accurately categorize stress levels with a potential accuracy exceeding 95%. |
|---|