Emotion state identification based on heart rate variability and genetic algorithm.

Area of Science:

• Biomedical Engineering
• Affective Computing
• Physiological Signal Processing

Background:

• Emotion recognition is crucial for human-computer interaction and mental health monitoring.
• Electrocardiogram (ECG) signals offer a non-invasive window into physiological states.
• Heart Rate Variability (HRV) analysis is a promising method for quantifying autonomic nervous system responses to emotions.

Purpose of the Study:

• To develop an effective, user-independent emotion recognition system using ECG.
• To differentiate between four emotional states: neutral, happiness, stress, and sadness.
• To investigate the efficacy of HRV features and machine learning for emotion classification.

Main Methods:

• Collected ECG data from ten male subjects exposed to visual and auditory stimuli.
• Extracted four categories of HRV features: time-domain, frequency-domain, Poincare plot, and differential.
• Employed a Support Vector Machine (SVM) classifier and a Genetic Algorithm (GA) for feature selection.

Main Results:

• Achieved a 90% emotion recognition rate with GA-based feature selection.
• Significantly improved recognition accuracy compared to systems without feature selection (52.2%).
• Demonstrated superior performance against existing user-independent emotion recognition systems.

Conclusions:

• The proposed ECG-based emotion recognition system effectively discriminates between four emotional states.
• GA-optimized HRV feature extraction enhances classification accuracy for emotion recognition.
• This user-independent approach shows significant potential for real-world affective computing applications.