Personalizing Chatbot Interactions for Mental Health Support

Problem Description

The development of effective chatbots for mental health support requires a nuanced understanding of therapeutic communication. Traditional methods in prompt engineering often fall short in providing the necessary customization for sensitive interactions with individuals experiencing depression. This thesis proposes the development of a prototype platform that personalizes interactions with a chatbot based on generative large language models (LLMs). The platform will enable experts to label real user conversations, using these insights to refine and optimize prompts for more empathetic and supportive interactions in mental health contexts.

Goal of the Thesis

The aim of this thesis is to develop and extend a platform that enhances personalized conversations between users and a chatbot designed to support individuals with depression. This user-centric approach in prompt engineering will leverage expert feedback to iteratively improve the chatbot’s interactions. The success of the platform will be judged by its ability to facilitate more effective, supportive, and satisfying conversations for users seeking mental health support.

Work Packages

  • Design and implement a platform to personalize chatbot conversations for mental health support, utilizing human-centered design principles (with a preference for Vue.js/React.js).
  • Conduct  (small scale) experimental study with real users and mental health experts to assess the effectiveness of the chatbot.
  • Analyze interaction data from experts and users to derive insights for continuous enhancement of the platform’s design and functionality.

Requirements

  • Proficiency in Python, Vue.js, or React.js.
  • Interest in generative AI, large language models, or human-computer interaction, particularly in the context of mental health.
  • Strong organizational and time management skills.
  • Proficiency in English.

Contact

If you are interested in applying or have any questions, please contact Leon Hanschmann (leon.hanschmann∂kit.edu). Submit a concise statement of your motivation, along with your CV and the most recent transcript of your academic records. 

Literature

Liu, June M., Donghao Li, He Cao, Tianhe Ren, Zeyi Liao, and Jiamin Wu. "ChatCounselor: A Large Language Models for Mental Health Support," 2023. https://doi.org/10.48550/arXiv.2309.15461.

Abd-alrazaq, Alaa A., Mohannad Alajlani, A. Alalwan, B. Bewick, Peter H Gardner, and M. Househ. “An overview of the features of chatbots in mental health: A scoping review,” 2019. https://doi.org/10.1016/j.ijmedinf.2019.103978.

Gupta, Vanshika, Varun Joshi, Akshat Jain, and Inakshi Garg. "Chatbot for Mental health support using NLP," 2023. https://doi.org/10.1109/INCET57972.2023.10170573.