Development of a Gemini LLM-Powered Mental Health Assistant for Early Screening in Indian Adolescents

DOI

10.2991/978-94-6239-756-9_18How to use a DOI?

Keywords

Adolescent Mental Health; Large Language Models; Gemini LLM; Chatbot; Conceptual Framework; Technical Architecture; Early Screening; India; School Health; FastAPI; Streamlit; adaptive assessment; conversational AI; LLM-driven questioning; prompt engineering

Abstract

Background

/Introduction: Adolescent mental health disorders are on the rise in India, where access to professional support is severely hampered by workforce shortages, stigma, and infrastructural barriers. Early identification within school settings is critical but seldom put into practice. Recent advances in large language models provide unprecedented opportunities for culturally sensitive, scalable mental health screening tools that can begin to fill these gaps, particularly in resource-constrained environments.

Objective

The objectives of the paper are to present the conceptual framework, ethical considerations, and technical architecture of the Gemini LLM-powered chatbot intended for the early screening of mental health issues in Indian adolescents. The work focuses on accessibility, ethical safeguards, multilingual support (English, Marathi, Assamese) and seamless integration within educational health programs.

Materials and Methods

/Research Design and Methods: A chatbot system is developed as a functional prototype using Python, FastAPI, and Streamlit, embedding Gemini LLM for conversational assessment based on youth-adapted versions of PHQ-9 and GAD-7 tools. Advanced functionalities include intelligent session management, AI-driven report generation, multi-layered emergency detection protocol, data anonymization, and accessibility enhancements. Most importantly, the system employs Gemini LLM for both writing reports and for extremely creative adaptive question generation work: the LLM decides or creates each following question on the fly, considering previous user answers, session state, and PHQ-9/GAD-7 coverage tracking. As a result, the prototype is no longer just a static survey with a summarizer but instead a conversational LLM-driven assessment. The methodology presents the design principles of the system and its technical implementation; further, preliminary technical validation is discussed and an agenda for future clinical evaluation is proposed.

Results

/Findings: The functional prototype illustrates a solid and effective technical base of the system, on average the API response time was 0.8 s and the simulated uptime was 99.8%. Besides that, the system that supports adaptive multilingual interactions and safety monitoring in real time, also reveals strong prototype-level indices of internal consistency across sessions. These results are indicative of the technological implementation and stability of the platform; however, at this phase, there are no clinical accuracy or diagnostic performance claims, and complete clinical validation will accompany a MINI-KID–based study in the future.

Conclusions

This technical prototype of the LLM-powered Screening Chatbot demonstrates much potential for scalable, effective, and ethical adolescent mental health screening in Indian schools. Its conceptual framework has successfully bridged the access and stigma barriers through techno-logical innovation and provided a blueprint for culturally sensitive early identification and an efficient allocation of resources.

Novelty/Originality: The paper contributes a detailed technical blueprint and ethical framework for leveraging Gemini LLM technology to screen adolescents, integrating region-specific languages and school health systems in India. Its open, modular, privacy-preserving architecture advances the field of responsible AI for public health.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Anujyoti Sonowal
AU  - Alaka Omprakash Chandak
PY  - 2026
DA  - 2026/08/31
TI  - Development of a Gemini LLM-Powered Mental Health Assistant for Early Screening in Indian Adolescents
BT  - Proceedings of the Conference on Bioengineering for Global Health (SYMRESEARCH 2.0 2025)
PB  - Atlantis Press
SP  - 253
EP  - 270
SN  - 2468-5747
UR  - https://doi.org/10.2991/978-94-6239-756-9_18
DO  - 10.2991/978-94-6239-756-9_18
ID  - Sonowal2026
ER  -