Research On Intelligent Inference Algorithm Based On Emergency Medicine

In the first aid scene,some injuries are highly lethal and disabling,and if they are not handled properly,they will have serious consequences for the patients.Therefore,the relevant research on emergency medicine has received unanimous attention at home and abroad.At the same time,due to the complexity of medical disciplines and the complexity of the first aid scene,it is difficult for the relevant medical staff to judge the injury condition at the first time.Therefore,this study uses the relevant technology of natural language processing to propose an emergency medical knowledge inference algorithm,which aims to help the medical staff in the emergency scene make the auxiliary decision of the injury condition quickly.Firstly,this study carries out PCA principal component analysis based on the relevant medical record data in the emergency database of 301 Hospital,and clearly divides the critical injury conditions of emergency medicine in combination with the experience of relevant emergency medical experts.At the same time,an ontology model based on the subject of main injury conditions is constructed.Based on the structure of the ontology model,the corresponding knowledge triple data are extracted from authoritative medical data such as《黄家驷外科学》,《内科学》,and the emergency medical knowledge graph is constructed by using neo4j technology,as well as the interactive environment of emergency medical knowledge base interacting with algorithm is constructed by using python.Secondly,this study selects the CCKS and CMID as public data sets and the data sets marked with docanno software as the training set of this inference algorithm.After analyzing the data storage structure of the data set,a series of techniques such as padding mask and text sequence sorting in batch are used to process the data set,and a standardized screening model of medical entities is proposed to filter the data in the data set.Then,this paper proposes an inference model based on natural language processing technology.By constructing three neural network models NET-N,NET-C and NET-D respectively,this paper infers the medical entity and text intention of the target medical text,and uses the pytorch deep learning framework to realize and experiment.By trying to use different pre training models as the word embedding layer for experiments,it is finally concluded that the entity recognition network NET-N adopts hfl/chinese-bertwwm-ext,the medical entity knowledge base matching network NET-C adopts hfl/chinese-bert-wwm-ext large,and the text intention inference model NET-D adopts hfl/chinese-bert-wwm-ext large.The final performance index of the three models is about85%.Finally,this paper integrates the proposed inference algorithm with the interactive environment of knowledge base,realizes the actual implementation of emergency medical inference algorithm,and tests the effect improvement of the algorithm in the form of question and answer system.Several typical test cases are selected for experiment and result analysis.Finally,it is concluded that the emergency medical inference algorithm can effectively improve the relevant results of medical inference problems.