Knowledge graph-driven lightweight medical large model system

Through the lightweight medical big model system driven by knowledge graph, the problems of medical big model in privacy protection, hallucination and knowledge update and iteration are solved, efficient and accurate medical diagnosis services are achieved, and the transparency and reliability of the model are enhanced.

CN120376117APending Publication Date: 2025-07-25XUZHOU MEDICAL UNIVERSITY
View PDF 6 Cites 1 Cited by

Patent Information

Application Number
CN202510863892.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-06-26
Publication Date
2025-07-25

AI Technical Summary

Technical Problem

Medical big models have challenges in privacy protection, hallucinations and instability, knowledge updates and iterations, and it is difficult to provide diagnostic services efficiently and accurately in the medical and health field.

Method used

A lightweight medical big model system driven by knowledge graph is adopted, including a question model, a summary model and extracting and verification summary model, collect patient information through multiple rounds of dialogue, combine medical knowledge graph for diagnosis, optimize model parameters and training methods, and ensure the transparency and accuracy of diagnosis.

Benefits of technology

It realizes efficient and accurate medical diagnosis, reduces hallucination problems, reduces update costs, improves the transparency of the model and user trust, and ensures the reliability and coverage of diagnostic results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120376117A_ABST
    Figure CN120376117A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of medical large models, and relates to a lightweight medical large model system driven by a knowledge graph. Comprising a questioning large model for collecting patient information through multiple rounds of dialogues, a summarizing model for comprehensively analyzing symptoms and medical history of a patient from the patient information and providing a preliminary diagnosis result and treatment suggestions, and a symptom entity and a disease entity from the patient information and the diagnosis result. Constructing a first candidate disease entity list by using the extracted disease entities, and constructing a second candidate disease entity list by acquiring related disease entities from a knowledge graph according to the extracted symptom entities, according to all the candidate disease entities, obtaining an extraction verification summary model of supplementary knowledge information of all the candidate disease entities from the knowledge graph; according to the method, the diagnosis process is perfected, the illusion problem is relieved, the interpretability is enhanced, and quick response and low-cost iteration are realized.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Training method of inquiry big model, inquiry method and device based on big model

    CN117747087A

  • Knowledge graph-driven medical large model diagnosis method

    CN118280562A

  • Disease prediction and risk assessment method based on large medical model

    CN118280570A

  • Interrogation method and system based on knowledge graph and generative large model

    CN117854748A

  • Intelligent diagnosis method and system based on knowledge graph and large model

    CN119339927A

Cited By

  • Internet hospital intelligent hospital guide method based on multi-modal information and double-model fusion

    CN121416042A