Analysis of Deep Learning based Emotion Charting Techniques for Parkinson’s Patients using Physiological Signal Analysis

Abstract

In psychology, the term "emotion recognition" describes the process of attributing emotional states based on the observation of nonverbal visual and aural clues. Perfect emotional exchanges between humans and computers can enhance communication. Emotional interactions are advantageous for various applications because they have a significant impact on cognitive functions of the human brain, such as learning, memory, perception, and problem-solving. It may also be applicable to contemporary healthcare, particularly in dealings with Parkinson’s disease sufferers. The second most prevalent neurodegenerative condition, Parkinson’s Disease (PD), impairs the ability to recognize and express emotions. Different emotion recognition systems are appropriate for various uses depending on the application domain. Nowadays, the concept of emotion recognition is extremely widespread. With the aid of IoT, physiological signals offer a suitable method to identify human emotion. There are several ways that emotions can be expressed, including through speech, behavior changes, facial expressions, and physiological markers. Physical signs provide a clearer understanding of emotion categorization. In order to construct a cutting-edge deep learning architecture for emotion charting for Parkinson’s disease, the associated parameters derived from the physiological signals, i.e. EEG, during emotion identification are investigated and evaluated in this study. Using this technique, one can readily forecast the victim’s emotional state when conducting an investigation or monitoring the health of Parkinson’s disease patients. In this thesis, we have proposed a deep learning based framework which can classify emotions of a PD patient using their EEG signatures. The results indicate that the framework can be improved to accurately classify emotions.