Preliminary pathogenesis diagnosis method based on knowledge map

Abstract

A preliminary pathogenesis diagnosis method based on a knowledge map includes the following steps: 1, collecting electronic medical records to construct a original dataset; 2, performing entity identification and relation extraction on the electronic medical records to construct an entity and relationship dataset in the RDF format; 3, constructing a knowledge map based on the above datasets; 4, constructing a symbiotic relationship model that predicts the relationship between diseases through the knowledge map; 5, preliminarily diagnosing the pathogenesis based on the symbiotic relationship model. The method, by using a deep learning algorithm represented by a convolutional neural network and a recurrent neural network, constructs the knowledge map, and uses the correlation of the medicalrecord data to extract high-level abstract attributes from the original information such as the medical record texts.

Description

technical field [0001] The present invention relates to data mining, knowledge map and deep neural network, in particular to a method for preliminary diagnosis of etiology based on knowledge map. Background technique [0002] Knowledge graph is one of the technologies of Semantic Web, which has become a research focus in the development of current search engine technology. Google is the advocate and pioneer of this concept, and expects to describe various entities and concepts in real time through the knowledge graph, as well as the relationship between them. The knowledge graph extracts knowledge from Internet texts, constructs a relational network in the form of a graph, and provides researchers with a "relational" perspective to analyze and research issues. [0003] As a big data technology, knowledge graph not only has the characteristics of visualization, but also facilitates the analysis of the relationship between entities. The knowledge map expresses the massive in...

AI Technical Summary

Problems solved by technology

[0007] Although there are already some systems for diagnosing causes based on electronic medical records, their core functions are still focused on rule-based review, and the level of intelligence is still low

[0008] In summary, there are many problems that need to be solved in the traditional algorithm for diagnosing etiology based on electronic medical records, including single diagnostic rules, high update cost, low frequency, and scattered medical record data. It is impossible to find abnormalities through big data, and the more common one is etiological diagnosis. The name of the patient's disease in the required medical records cannot match the symptoms, and the cause of the disease in the medical records cannot be initially diagnosed. There is no effective solution to these problems

Images