Research on the Medical Knowledge Deduction Based on the Semantic Relevance of Electronic Medical Record

Abstract

Abstract This paper studies the extraction of information from unstructured text data of medical literature and electronic medical records in the field of medicine, and proposes a TCM-KR method of knowledge reasoning based on electronic medical records to enhance association rules, and carries out a study on association characteristics in the field of the electronic medical record. This method abstracts the word bag representation mode of text semantics from the unstructured data representation and integrates the correlation information of the knowledge graph of the medicine domain. The method based on a graph convolutional network was used to predict the unknown associations' relations between viscera, channel tropism, and channel distribution. The experimental results show that the TCM-KR method can efficiently infer a large amount of high-quality triple knowledge from the unstructured text data of medicine, and predict the correlation characteristics of Syndromes-Viscera, Chinese medicinal-Channel tropism, Acupoints-Channel distribution in treating lumbar intervertebral disc prolapse and provide a dedicated machine learning model and guidance for clinical diagnosis and treatment.

Keywords

• Knowledge graph
• Semantic relevance
• Text semantic enhancement
• Deep learning
• Electronic medical record