Mental Health Assessment Using EEG Sensor and Machine Learning

This research explores a new technique for classifying human emotions based on EEG signals, focusing on improving accuracy. The method utilizes a unique combined classifier that leverages features extracted from both time domain analysis and Discrete Wavelet Transform (DWT), aiming to achieve a superior performance level. This project presents two innovative advancements: as a fresh feature, an original variable is introduced along with a combined classifier that integrates SVM and HMM algorithms. The results reveal a 5% and 1.5% accuracy boost on the valence and arousal axes, respectively, achieved through the combined features. The combined classifier further enhances accuracy by 3% compared to the SVM classifier alone. One key application for a highly accurate emotion classification system lies in offering a robust tool for psychologists to diagnose mental illnesses related to emotions, the system developed in this project holds considerable potential in fulfilling this essential function.