A Cognitive IoMT Approach for Stress Level Detection using Hyper FusionXG

Stress is a prevalent health concern affecting individuals across various age groups, contributing to both psychological and physiological disorders. Existing stress detection methods often suffer from limitations such as dependency on complex physiological signals like electrodermal activity (EDA) and sweat rate, which may not be accessible or practical for individuals with disabilities. Moreover, many approaches rely on limited datasets or fail to achieve high accuracy in real-world scenarios. To address these challenges, this study introduces HyperFusionXG, an Internet of Medical Things (IoMT)-driven framework that integrates electroencephalogram (EEG) and electrocardiogram (ECG) signals for enhanced stress detection. Unlike conventional methods, this approach reduces reliance on additional physiological signals while ensuring inclusivity and scalability. By leveraging advanced machine learning (ML) techniques, including feature selection, hyperparameter tuning, and model optimization, HyperFusionXG achieves an accuracy of 99.3%, demonstrating its robustness in stress assessment. The framework employs early fusion of EEG and ECG signals, followed by feature refinement using Recursive Feature Elimination (RFE) and classification through an optimized XGBoost model. The proposed method not only ensures non-invasive and continuous monitoring but also captures intricate stress-related patterns effectively, making it a reliable solution for real-world stress detection, particularly for individuals with disabilities. This work underscores the transformative potential of IoMT-driven machine learning in advancing stress monitoring technologies, paving the way for more accessible and efficient healthcare solutions.