Knowledge graph-based old medicine new use analysis method and system

Abstract

The invention discloses a knowledge graph-based old medicine new use analysis method. The method comprises the steps of obtaining a knowledge graph; extracting all disease entities, medicine entitiesand gene entities and relationships among the disease entities, the medicine entities and the gene entities in the knowledge graph; vectorizing all entities and relationships to obtain a feature vector of each entity and a feature vector of the relationship; determining one or more candidate machine learning models; training the candidate machine learning model; verifying and comparing the candidate machine learning models, and selecting an optimal machine learning model; and predicting a specific disease by applying the optimal model to obtain a plurality of medicine entities related to the specific disease.

Description

technical field [0001] The invention relates to the fields of chemical informatics and bioinformatics. Specifically, the present invention relates to a method and system for analyzing new uses of old drugs based on knowledge graphs. Background technique [0002] The biomedical knowledge graph is usually a multi-relational directed graph, in which nodes represent entities and edges represent relationships. The knowledge base of this graph structure has become a resource with great value and potential application value, which is helpful for precision medicine. and clinical decision support, which play an increasingly important role in healthcare practice and research. However, due to the usually large and noisy knowledge graphs constructed from biomedical data and the complexity and high knowledge requirements of downstream tasks, there are still many challenges in construction and application. [0003] The task of link prediction between entities is currently the most widel...

AI Technical Summary

Problems solved by technology

Therefore, although they have been successfully applied to knowledge graph tasks such as question answering, information extraction, and named entity disambiguation outside the biomedical domain, their use in biomedical knowledge graph applications still lacks experimental validation.

Images