EEG-Based Stress Level Detection Using Machine Learning

This project introduces a practical and intelligent approach to stress detection using EEG signals and
machine learning. By capturing and analyzing brainwave activity, the system identifies five key frequency
bands—delta, theta, alpha, beta, and gamma—through the use of a Butterworth bandpass filter. From these
bands, meaningful features are extracted to reflect the brain’s cognitive and emotional states. These features
are then processed through a hybrid pipeline combining unsupervised clustering with supervised learning
models, including Support Vector Machine (SVM), Random Forest (RF), and XGBoost, to classify stress
levels into Low, Moderate, and High. To bring this solution closer to real-world application, the trained
models are integrated into a user-friendly interface built with Streamlit, enabling real-time monitoring and
predictions. The result is a robust, automated system capable of supporting mental health assessments in
various settings—from clinical environments to personal wellness applications—offering both scalability
and accessibility.