A Trustworthy Medication AI Clinician by Multi-aspect RAGs

KID (Knowledge and Information Discovery) lab 的研究計畫 「A Trustworthy Medication AI Clinician by Multi-aspect RAGs」,旨在結合尖端人工智慧技術與可靠的醫療知識,開發一個可信賴的 AI 藥物臨床輔助系統。該系統核心採用多層面檢索增強生成模型(Multi-aspect Retrieval-Augmented Generators, RAGs),以提供精準且個性化的用藥建議,同時提升臨床決策的透明度與安全性。


該系統的主要功能包括提供患者專屬的藥物使用建議、監控藥物相互作用、提示潛在的不良反應以及動態更新最新的醫療知識。此外,透過結合數據可視化技術,醫護人員可以直觀地觀察建議的適用範圍及其可能影響,為臨床決策提供可靠的參考依據。


這項計畫的最終目標,是為醫療環境中日益增長的決策需求提供技術支持,讓 AI 不僅僅是輔助工具,更成為提升醫療品質和患者安全的重要夥伴。

Members

    國立成功大學電機系 Jen-Wei Huang (黃仁暐) 教授
    Hsin Yang, Hui-Hsin Xue, Cong-Yuan Dai, Yang-Kuang Lin
    Yu-Yun Mo, Li-Yuan Hung, Shang-Chun (Luke) Lu

    National Chen Kung University (NCKU), Taiwan. 2024.

Overall Architecture
Introduction to Our Model
Chat with Our Model
Other Demo Videos
This video demonstrate 3 questions about the drug Lipitor, and the answers shows that the model is able to answer the question with different information retrieved from different RAGs.
This video use 3 questions from the dataset of ChatDoctor, which shows that the model has the ability to answer the questions with professional knowledge.