Semi-supervised learning method for constructing medical knowledge graph from Chinese electronic medical records

Abstract

The invention relates to the technical field of Chinese electronic medical record processing, and discloses a semi-supervised learning method for constructing a medical knowledge graph from Chinese electronic medical records. The method is an end-to-end method, and comprises the following steps: firstly, obtaining an original corpus of a Chinese electronic medical record, performing operations such as data preprocessing, knowledge description system design, manual entity relationship marking and the like, and sorting related medical terms to construct a Chinese medical word segmentation dictionary; completing the knowledge extraction work in combination with a machine learning algorithm and a deep neural network; expanding a Chinese electronic medical record mark data set step by step in combination with a semi-supervised learning method, storing all extracted knowledge triples in a Neo4j database, and constructing a medical knowledge graph. According to the method provided by the invention, the Chinese electronic medical record annotation corpus is provided, and the medical knowledge in the medical record can be accurately identified.

Description

technical field [0001] The invention relates to the technical field of Chinese electronic medical records processing, in particular to a semi-supervised learning method for constructing a medical knowledge graph from Chinese electronic medical records. Background technique [0002] Knowledge Graph is essentially a language network, its nodes represent entities, and connections represent various semantic relationships between entities, which can connect scattered knowledge to each other to form a A huge, networked knowledge system built with the "semantic network" as the skeleton. As more and more semantic WWW data is opened on the Internet, various Internet search engine companies at home and abroad have begun to build knowledge graphs based on this to improve service quality, such as Google Knowledge Graph (Google Knowledge Graph), Baidu "Zhixin" and so on. The construction of knowledge graphs in the medical field is a major research hotspot at present. Electronic medical...

AI Technical Summary

Problems solved by technology

[0004] 1. Due to the inclusion of patient sensitive information, and the labeling of medical entities and entity relationships relies on domain knowledge, there is currently a lack of high-quality, large-scale Chinese electronic medical record labeling corpus in China

[0005] 2. Chinese electronic medical records contain a large number of professional medical vocabulary and special characters, and there is no clear boundary between Chinese words and words, resulting in very complex sentence features that are difficult to extract. Existing technologies cannot accurately identify medical knowledge in medical records

Images