Mental health chatbots : empathetic support by leveraging large language models | |
| Author | Kuragayala, Aakash |
| Call Number | AIT RSPR no.DSAI-25-02 |
| Subject(s) | Mental health services--Technological innovations Artificial intelligence--Medical applications Natural language generation (Computer science) Chatbots |
| Note | A research study submitted in partial fulfillment of the requirements for the degree of Master of Science in Data Science and Artificial Intelligence |
| Publisher | Asian Institute of Technology |
| Abstract | This study aims to develop a mental health chatbot that provides empathetic, evidence based support for individuals experiencing depression, frustration, low self-esteem, and loneliness by leveraging advanced large language models (LLMs). The research will focus on fine-tuning state-of-the-art LLMs using diverse mental health conversational datasets to personalize responses, ensuring they are compassionate and contextually relevant. By harnessing the capabilities of modern LLMs, the chatbot will address the growing need for accessible mental health support, offering a scalable solution that complements traditional therapeutic interventions while prioritizing user emotional well-being.To enhance the chatbot’s performance, the study will integrate Chain of Thought (COT) prompting combined with a therapeutic knowledge base, utilizing Retrieval-Augmented Generation (RAG) to improve response quality and transparency. COT will guide the chatbot to reason systematically, identifying emotions and suggesting appropriate coping strategies, while RAG will incorporate external resources, such as cognitive behavioral therapy (CBT) guidelines, to ensure responses are grounded in evidence-based practices. This approach aims to increase the chatbot’s ability to deliver coherent, empathetic, and therapeutically informed responses, addressing limitations in existing conversational AI systems for mental health applications.The research will rigorously evaluate the chatbot’s effectiveness by comparing the performance of the baseline model against the COT-enhanced version through comprehensive metric analysis. Quantitative metrics, such as ROUGE, BLEU, and METEOR, will assess linguistic accuracy and coherence, while qualitative measures, including empathy and safety evaluations, will ensure alignment with mental health needs. The study will also prioritize ethical considerations, incorporating privacy safeguards and crisis detection mechanisms to ensure safe and responsible deployment. By demonstrating improvements in response quality and user support, this research seeks to advance the development of AI-driven mental health tools, contributing to accessible and empathetic emotional care. |
| Year | 2025 |
| Type | Research Study Project Report (RSPR) |
| School | School of Engineering and Technology |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Data Science and Artificial Intelligence (DSAI) |
| Chairperson(s) | Chantri Polprasert; |
| Examination Committee(s) | Tripathi, Nitin Kumar;Chaklam Silpasuwanchai;Ratchainant Thammasudjarit; |
| Scholarship Donor(s) | AIT Scholarship; |
| Degree | Research Studies Project Report (M. Sc.) - Asian Institute of Technology, 2025 |