Research Areas: Artificial Intelligence, Machine Learning, ECG Signal Processing, Physiological Computing, Stress Detection, Healthcare AI

Project Description:
Psychological and physiological stress can influence cardiovascular activity and produce measurable changes in electrocardiogram (ECG) signals. This project investigates how ECG-derived biomarkers can be used with machine learning to analyze and classify different levels of stress.

Students will preprocess ECG signals, detect R-peaks, calculate heart rate, and derive heart-rate-variability (HRV) features. Both time-domain and frequency-domain HRV measurements can be explored, along with morphological ECG characteristics where appropriate.

Machine-learning models such as Logistic Regression, Support Vector Machines, Random Forest, XGBoost, and neural networks will be evaluated for distinguishing relaxed and stressed physiological states. Students may also study personalized models, subject-independent validation, and explainable-AI techniques to determine which cardiovascular features contribute most strongly to stress predictions.

Research Questions: How accurately can ECG and HRV features identify physiological stress? Which HRV features are most informative? Do personalized models perform differently from subject-independent models? Can machine-learning explanations reveal physiologically meaningful stress indicators?

Student Activities: ECG filtering, R-peak detection, HRV analysis, feature engineering, statistical analysis, machine-learning development, cross-validation, explainable AI, visualization, and research-paper preparation.

Technologies: Python, NeuroKit2, WFDB, NumPy, SciPy, Pandas, Scikit-learn, XGBoost, SHAP, Matplotlib, Jupyter/Google Colab.

Expected Outcomes: ECG/HRV signal-processing pipeline, stress-classification model, feature-importance analysis, student research presentation/poster, and potential conference publication.