Auxiliary diagnosis system for large-scale medical records based on deep learning

Abstract

The invention discloses an auxiliary diagnosis system for large-scale medical records based on deep learning. The auxiliary diagnosis system comprises a cyclic neural network model, a convolutional neural network model and a fusion calculating unit. The cyclic neural network model is obtained through training the cyclic neural network based on large-scale previous medical records with a diagnosisconclusion. The convolutional neural network model is obtained through training the convolutional neural network through the artificially generated medical records, wherein the artificially generatedmedical records are based on a correspondence between diseases and symptoms in a knowledge graph and obtained through arrangement combination of disease incidence rate and symptom incidence rate. Information of gender, age, main complaint, existing disease history, previous history, personal history, family history, physique examination and auxiliary examination is input into the cyclic neural network model and the convolutional neural network model. The fusion calculating unit generates diagnosis prompt which is related with the medical record according to biological calculating results of the cyclic neural network model and the convolutional neural network model.

Description

technical field [0001] The present invention relates to deep learning technology and natural language processing technology, in particular to an auxiliary diagnosis system based on large-scale medical record data. Background technique [0002] In the mid-1970s, Stanford University developed the world's first clinical decision support system CDSS (MYCIN). Then there are many CDSS based on knowledge base, which consists of three important parts: database (Data Repository), reasoning engine (Rules Engine) and human-computer interaction interface (Interface), usually using IF-THEN rules, or based on prior probability and Bayesian Statistics for Conditional Probability. Based on this, the huge and reliable clinical knowledge base has become an industry barrier for CDSS. With the development of technology, CDSS based on artificial intelligence technology has emerged in recent years. Through deep learning technology, the computer can learn past experience or clinical routine patt...

AI Technical Summary

Problems solved by technology

Research on decision-making methods for auxiliary diagnosis based on Chinese electronic medical records is still in its infancy. In China, Zhou Zhihua, Jiang Yuan and others used machine learning models for disease prediction and analysis, but they are all single-disease disease prediction models, which are difficult to directly apply to multiple features. In the prediction of multi-category medical data sets

Images