@inproceedings{hsu-etal-2025-medplan,
    title = "{M}ed{P}lan: A Two-Stage {RAG}-Based System for Personalized Medical Plan Generation",
    author = "Hsu, Hsin-Ling  and
      Dao, Cong-Tinh  and
      Wang, Luning  and
      Shuai, Zitao  and
      Phan, Thao Nguyen Minh  and
      Ding, Jun-En  and
      Liao, Chun-Chieh  and
      Hu, Pengfei  and
      Han, Xiaoxue  and
      Hsu, Chih-Ho  and
      Luo, Dongsheng  and
      Peng, Wen-Chih  and
      Liu, Feng  and
      Hung, Fang-Ming  and
      Wu, Chenwei",
    editor = "Rehm, Georg  and
      Li, Yunyao",
    booktitle = "Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 6: Industry Track)",
    month = jul,
    year = "2025",
    address = "Vienna, Austria",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/display_plenaries/2025.acl-industry.76/",
    pages = "1072--1082",
    ISBN = "979-8-89176-288-6",
    abstract = "Despite recent success in applying large language models (LLMs) to electronic health records (EHR), most systems focus primarily on assessment rather than treatment planning. We identify three critical limitations in current approaches: they generate treatment plans in a single pass rather than following the sequential reasoning process used by clinicians; they rarely incorporate patient-specific historical context; and they fail to effectively distinguish between subjective and objective clinical information. Motivated by the SOAP methodology (Subjective, Objective, Assessment, Plan), we introduce MedPlan, a novel framework that structures LLM reasoning to align with real-life clinician workflows. Our approach employs a two-stage architecture that first generates a clinical assessment based on patient symptoms and objective data, then formulates a structured treatment plan informed by this assessment and enriched with patient-specific information through retrieval-augmented generation. Comprehensive evaluation demonstrates that our method significantly outperforms baseline approaches in both assessment accuracy and treatment plan quality. Our demo system and code are available at https://github.com/JustinHsu1019/MedPlan."
}