Super-indication drug analysis system fusing knowledge graph and large language model

By integrating knowledge graphs and large language models into an off-label drug use analysis system, high-precision and interpretable off-label drug identification is achieved, solving the problem of low accuracy in existing technologies and providing a rigorous and flexible intelligent solution.

CN122369786APending Publication Date: 2026-07-10PEKING UNIVERSITY THIRD HOSPITAL (THE THIRD CLINICAL MEDICAL SCHOOL OF PEKING UNIVERSITY) +2
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
PEKING UNIVERSITY THIRD HOSPITAL (THE THIRD CLINICAL MEDICAL SCHOOL OF PEKING UNIVERSITY)
Filing Date
2026-06-04
Publication Date
2026-07-10

AI Technical Summary

Technical Problem

Existing off-label drug detection technologies struggle to balance medical rules with semantic flexibility, lack effective support for long-tail entities and dynamic knowledge, resulting in low accuracy and a lack of interpretability.

Method used

The off-label drug use analysis system, which integrates knowledge graphs and large language models, achieves high-precision entity recognition and deterministic logical matching by combining a four-level cascaded matching strategy and a medical knowledge graph through entity recognition and standardization modules, knowledge enhancement modules, symbolic knowledge layers, and neural reasoning layers. It is further supplemented by mechanism similarity analysis and clinical evidence evaluation using large language models.

Benefits of technology

It significantly improves the accuracy and interpretability of off-label indications, reduces the risk of hallucinations, and provides an intelligent solution that is both rigorous and flexible, supporting clinical decision-making and drug regulation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122369786A_ABST
    Figure CN122369786A_ABST
Patent Text Reader

Abstract

This invention relates to an off-label drug use analysis system integrating knowledge graphs and large language models, belonging to the field of medical information processing technology. It solves the problems of existing off-label drug detection systems, which struggle to balance medical rules and semantic flexibility, while lacking effective support for long-tail entities, dynamic knowledge, and interpretable reasoning. The system includes: an entity recognition and standardization module for semantic parsing of patient medical texts to obtain standardized entities; a knowledge enhancement module for retrieving corresponding medical knowledge from a standard medical knowledge graph based on standardized entities; a symbolic knowledge layer for matching standardized entities based on medical knowledge to obtain initial off-label drug use judgment results; a neural reasoning layer for performing similarity analysis based on medical knowledge to obtain auxiliary off-label drug use analysis results; and a fusion decision module for generating the final off-label drug use judgment result based on the initial off-label drug use judgment result and the auxiliary off-label drug use analysis results. This system enables off-label drug use analysis.
Need to check novelty before this filing date? Find Prior Art