1479-P: GK Charger: An LLM-Empowered Knowledge Platform for Interactive Education in Glucokinase-Based Diabetes Therapy
Diabetes · 05.06.2026 · 75 (Supplement_1)
DOI: 10.2337/db26-1479-p ↗Kurz gefasst
Fachartikel in der Zeitschrift Diabetes, erschienen am 05.06.2026. Verfasst von LINGGE FENG und 7 weiteren Autor:innen.
Themen: Diabetes mellitus
Abstract (Originalsprache)
Introduction and Objective: Glucokinase (GK) is a central regulator of glucose homeostasis, with dysfunction closely linked to T2DM. As GK activators enter clinical practice, efficient and accurate knowledge transfer is required to support clinical decision-making and individualized therapy. However, GK biology spans complex multi-organ networks (pancreas-liver, gut-pancreas, and pancreas-brain), exceeding the capacity of conventional manual curation.
Methods
We developed GK Charger, an intelligent interactive platform integrating large language models (LLMs) with a structured knowledge base (GK Brain). The system includes: (1) a curated, modular GK knowledge base derived from literature, clinical guidelines, expert consensus, and drug information; (2) an LLM-enabled knowledge graph capturing entities and semantic relationships; and (3) attention-based embedding models enabling semantic association, reasoning, and identification of users’ knowledge gaps. Natural language interaction supports physician education, pharmaceutical training, and patient consultation. All outputs undergo periodic expert review to ensure scientific accuracy.
Results
Pilot implementation demonstrated enhanced clinical teams’ efficiency and depth of understanding of GK-mediated glucose regulation. Literature-linked outputs improved scientific rigor, while semantic simplification translated complex mechanisms into patient-comprehensible information, supporting clinical management adherence. User interaction-driven learning enabled automated identification of knowledge gaps and iterative optimization of the knowledge base, establishing a closed-loop knowledge management system.
Conclusion
GK Charger illustrates a scalable and reproducible paradigm integrating artificial intelligence with authoritative medical knowledge, supporting precision education and digital healthcare delivery in T2DM. Disclosure L. Feng: None. T. Wu: None. Q. Guo: None. L. Liu: None. C. Li: None. F. Tang: None. Y. Zhang: None. L. Chen: None.
Zitieren
APA
FENG, L., WU, T., GUO, Q., LIU, L., LI, C., TANG, F., ZHANG, Y., & CHEN, L. (2026). 1479-P: GK Charger: An LLM-Empowered Knowledge Platform for Interactive Education in Glucokinase-Based Diabetes Therapy. Diabetes, 75(Supplement_1). https://doi.org/10.2337/db26-1479-p
Vancouver
FENG L, WU T, GUO Q, LIU L, LI C, TANG F, et al. 1479-P: GK Charger: An LLM-Empowered Knowledge Platform for Interactive Education in Glucokinase-Based Diabetes Therapy. Diabetes. 2026;75(Supplement_1). doi:10.2337/db26-1479-p
BibTeX
@article{FENG2026,
author = {FENG, LINGGE and WU, TUNG-LIN and GUO, QUANLE and LIU, LIYUAN and LI, CHANGHONG and TANG, FUXING and ZHANG, YI and CHEN, LI},
title = {1479-P: GK Charger: An LLM-Empowered Knowledge Platform for Interactive Education in Glucokinase-Based Diabetes Therapy},
journal = {Diabetes},
year = {2026},
volume = {75},
number = {Supplement_1},
doi = {10.2337/db26-1479-p}
}
EUVIMED