A Patient Digital Model Retrieval Method and System Based on Event Graph
By constructing a patient digital model based on event graphs and utilizing a large language model and dynamic weight adjustment mechanism, the problem of insufficient utilization of unstructured data in existing technologies is solved, enabling efficient and accurate patient digital model retrieval and personalized diagnosis and treatment support.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-02-14
- Publication Date
- 2026-05-26
AI Technical Summary
Existing technologies struggle to fully leverage the value of unstructured or semi-structured medical data when building digital patient models, resulting in inaccurate information matching, an inability to accurately reflect the characteristics of a patient's condition, and a lack of dynamic representation capabilities, making it difficult to support personalized treatment decisions.
An event graph-based approach is adopted to construct a timeline event graph by intelligently parsing electronic medical records, extract key diagnosis and treatment events using a large language model, and perform patient digital model retrieval by combining a dynamic weight adjustment mechanism, thereby realizing information association across time nodes and multimodal data fusion.
It significantly improves the accuracy and clinical relevance of patient digital models, enabling rapid matching of similar cases, supporting the development of personalized treatment plans, improving diagnostic and treatment efficiency and precision, and promoting the development of precision medicine.
Smart Images

Figure CN122087083A_ABST