Home // HEALTHINFO 2025, The Tenth International Conference on Informatics and Assistive Technologies for Health-Care, Medical Support and Wellbeing // View article

School Health Dialogue: A Prompt-Expansion and Response-Visualization Framework

Authors:
Hayato Tomisu
Kazue Yamamura
Junya Ueda
Tsukasa Yamanaka

Keywords: large language models; decision graph visualization; conversation management; adolescent wellness; school infirmary.

Abstract:
Adolescents often express mental or physical discomfort in vague terms, thereby placing a high cognitive burden on school nurses, who must interpret incomplete information. This study proposes a two-layer framework to improve school health communication by transforming ambiguous student utterances into structured, explainable dialogue flows. The first layer, auto-prompt expansion, enriches student input into slot-based representations. The second, the Prompt-Graph domain-specific language, maps these representations onto a transparent decision graph for nurse supervision. The system integrates large language model orchestration, animated avatars, and real-time graph rendering. In an evaluation of 50 student complaints, prompt expansion achieved an F1 score of 0.82, whereas slot extraction scored 0.43 owing to lexical variability. AI-based rubric evaluations yielded high tone scores, indicating consistent empathy in responses; however, lower ratings for accuracy and completeness revealed deficiencies in medical specificity and follow-up guidance. Future studies will address clinical tuning, symptom normalization, and long-term field validation.

Pages: 29 to 34

Copyright: Copyright (c) IARIA, 2025

Publication date: October 26, 2025

Published in: conference

ISSN: 2519-8491

ISBN: 978-1-68558-312-5

Location: Barcelona, Spain

Dates: from October 26, 2025 to October 30, 2025