Towards Emotion-Sensitive Conversational User Interfaces in Healthcare Applications

Area of Science:

• Natural Language Processing
• Artificial Intelligence in Healthcare
• Human-Computer Interaction

Background:

• Conversational agents are increasingly used in mobile health (mHealth) applications.
• Effective emotion perception and response are crucial for successful human-agent interaction, especially in healthcare.
• Accurate emotion detection in healthcare is complex due to contextual and medical factors.

Purpose of the Study:

• To introduce a concept for analyzing emotions and sentiments in mHealth conversational interfaces.
• To enhance conversational agents' ability to understand and respond to user emotions within a healthcare context.

Main Methods:

• Utilized bot technology (Synthetic Intelligence Markup Language) and deep learning for emotion analysis.
• Employed treebank annotation and recursive neural networks for classifying sentiments/emotions into seven categories and three strength levels.
• Integrated classification results into chatbot response selection logic.

Main Results:

• Developed a novel approach for emotion and sentiment analysis in mHealth conversational agents.
• Demonstrated the classification of emotions with seven categories and three strength levels.
• Enabled chatbots to select appropriate responses based on analyzed user emotions.

Conclusions:

• The proposed approach enhances conversational agents' emotion-sensitivity in mHealth applications.
• This method allows for more appropriate and empathetic user responses, better addressing user concerns.
• The concept is applicable across various mHealth use cases to improve user experience and care.