Predicting Emotional States from Wearable ECG

This study aims to enhance the accuracy of emotion recognition systems using wearable electrocardiogram (ECG) data. Leveraging the DREAMER dataset, which comprises ECG recordings and self-reported emotional states, we focus on the feature engineering process to identify the most informative physiological features. Various machine learning models, including random forests, support vector machines (SVM), and logistic regression, are trained to classify emotions across the dimensions of valence, arousal, and dominance. Our findings demonstrate that random forests, particularly when applied to carefully selected ECG features, perform well in emotion classification. These results underscore the importance of feature selection in emotion recognition, paving the way for more personalized and real-time applications in mental health monitoring and human-computer interaction.