Automated Classification of Happy and Sad Emotional States from PPG Signal Using Time Domain Analysis

Automated emotion detection and analysis have become one of the most important domains in recent years because of its applicability in health care, education, entertainment industry, robotics, and marketing sectors. Emotion recognition gives an accurate estimation of the mental condition of a person and indicates the inherent activity or thinking state of the mind. In the field of emotion recognition, artificial intelligence-based analysis has become an essential part of research along with the domains of medical science, cognitive science, computer science, and neuroscience. Emotion of any person can be judged from the gesture, face recognition, and body movements. But these are not conclusive enough because of the ability of a person to suppress these responses at will. The present paper presents a computationally simple approach to estimate and classify the emotional states of happy and sad based on a single feature calculated from a single lead PPG signal available in DEAP dataset. Classification is carried out using a binary classification rule and does not require any complex classifier. Results show the justification of the proposed feature and the validity of the approach. The present method may be applied in real-time systems for the detection of sadness or mental state of any individual owing to its simplicity in operation and implementation.