1. Motivation and Problem Statement

1.1 Background and Motivation

The integration of electroencephalography (EEG) and electrocardiography (ECG) signals in the realm of emotion recognition is a growing area of interest, bridging neuroscience, machine learning, and personalized health applications. Despite this, few resources comprehensively cover both the breadth of applications and the array of models available for EEG-ECG integration, leaving a gap for researchers and practitioners who seek a unified resource. Recognizing this need, this project provides a foundational resource with literature review and implementation framework that spans both applications of EEG-ECG in emotion recognition and an extensive survey of machine learning and deep learning models tailored to this domain.

Coming from a non-STEM background, I was motivated to approach this project not only as a technical contribution but also as a personal learning journey. By synthesizing existing applications—from biofeedback and neurocardiology to adaptive gaming and hybrid brain-computer interfaces—and models, including decision trees, convolutional neural networks (CNNs), temporal convolutional networks (TCNs), graph neural networks (GNNs), and generative models, I aim to make this field more approachable for others like myself who are entering from diverse academic and professional backgrounds.

The practical implementation of a stress detection system provides an accessible model that can be reproduced and adapted for further research. Together, these elements establish this project as a comprehensive, foundational resource that supports continued innovation and interdisciplinary research in EEG-ECG emotion recognition. By doing so, it establishes a comprehensive reference point that informs future research and supports diverse audiences, from experienced technical researchers to interdisciplinary professionals entering this field.

1.2 Problem Statement

Research on heart-brain dynamics through EEG and ECG signals covers a diverse range of applications—from emotion recognition to adaptive gaming and health monitoring. However, these studies are often fragmented, focusing on specific applications or particular model architectures. As a result, comprehensive resources that bridge applications and model techniques with a practical implementation framework are scarce. This gap limits the accessibility of the field, making it challenging for new researchers to enter and for experienced professionals to extend their knowledge across interdisciplinary applications.

This project addresses this gap by delivering a foundational resource: (1) a structured review of applications in heart-brain dynamics; (2) a detailed review of machine learning and deep learning models, particularly for emotion recognition, including those suitable for EEG-ECG data; and (3) a practical implementation of stress detection as a focused use case. This implementation provides an accessible model that can be readily reproduced and adapted for further research. By combining these elements, this resource aims to support a wide audience in understanding, comparing, and applying EEG-ECG methods for emotion recognition and related studies.

1.3 Objectives

The primary objective of this project is to provide a comprehensive and foundational resource that establishes a baseline for EEG-ECG emotion recognition and using stress detection as a focused use case. To achieve this, we focus on three main goals:

1. Comprehensive Survey of Applications in Heart-Brain Dynamics: Present a structured overview of EEG-ECG applications across fields like biofeedback, personalized health monitoring, neurocardiology, and adaptive gaming. This review contextualizes emotion recognition within a broad range of applications, providing a practical guide for researchers exploring the interdisciplinary potential of EEG-ECG studies.

2. Extensive Model Review for Emotion Recognition: Offer a detailed survey of machine learning and deep learning models for EEG-ECG-based emotion recognition, spanning traditional approaches (e.g., decision trees, SVMs) to advanced architectures (e.g., CNNs, LSTMs, GANs). This review aids researchers in understanding model strengths, weaknesses, and applicability for studies involving emotion and stress.

3. Practical Stress Detection System: Develop a reproducible, machine learning-based system for binary stress classification using EEG and ECG data. This accessible model provides a practical baseline that can be adapted for future research, demonstrating a hands-on application of stress detection methods in heart-brain dynamics.

These objectives establish this project as a foundational resource for EEG-ECG emotion recognition, offering a reference point for future research and interdisciplinary exploration in heart-brain dynamics.