Research On Named Entity Recognition Of Chinese Electronic Medical Records

With China’s rapid economic development and the growing need for a better life,people pay more attention to health care and want to enjoy more intelligent and reliable medical services,so that they can have a more comprehensive understanding of their own health status.However,the problem of uneven and inadequate development in the medical field still exists,and the construction of information technology in the medical field is still in the development stage.The use of natural language processing technology such as named entity recognition to intelligently analyze and excavate the text of electronic medical records can help establish a sound and intelligent electronic medical record system,assist medical personnel in making diagnostic decisions and promote the construction of information technology in the medical field.The earliest approach to the main methods of named entity recognition research was the use of the Hidden Markov Model for English named entity recognition,followed by the emergence of rule-based,traditional machine learning methods.In the field of named entity recognition,neural network models have been shown to improve the accuracy of named entity recognition,reduce the cost of model construction,and provide a significant improvement in model performance.Therefore,the neural network model has broad application prospects and it is necessary to explore its performance in the electronic medical record naming entity recognition model.Named entities in electronic medical records are usually medical terminology,most of which are low-frequency words in the corpus.At the same time,electronic medical records contain patient privacy information,making it difficult to disclose electronic medical record data and lacking a corpus of electronic medical record markup.Compared with English electronic medical record,Chinese electronic medical record named entity identification research faces more challenges,such as boundary definition of Chinese words,ambiguity of lexical meaning and syntactic ambiguity problems,etc.In response to the above issues,this paper investigates the identification of naming entities in Chinese electronic medical records.A fine-grained semantic representation learning algorithm for Chinese characters is proposed to improve the recognition performance of the named entities of the model by learning semantic information of the Chinese character paraphernalia and prefix components.A weak domain migration algorithm is proposed,which successfully achieves cross-domain feature and parameter migration and solves the problem of lack of Chinese electronic medical records corpus.The model proposed in this paper was experimentally validated,and the identification of Chinese electronic medical record named entities was investigated.The main lines of work of this paper are as follows.1.A Chinese electronic medical record named entity recognition model was constructed based on the attention mechanism.Think of the named entity recognition task as a character-level sequence annotation task to solve the problem of infinity between Chinese words and possible errors in the participle.Establish dependencies between any two words in a sentence by introducing attention mechanisms.An algorithm for learning the fine-grained semantic representation of Chinese characters is proposed.Enhance the semantic representation of text sequences in electronic medical records by learning character feature semantics,prefix feature semantics,and lexical feature semantics.2.In response to the common lack of annotated corpus in electronic medical records,a transfer learning framework was introduced in the Chinese electronic medical record named entity identification with out-of-domain data,and a weak domain migration algorithm was proposed.The algorithm can reduce the feature distribution differences between domains,achieve feature and parameter migration between source and target domains,and solve the problem of lack of annotation corpus in target domains.3.Experiments were conducted using real electronic medical record data using the model presented in this paper.The experimental results show that the model proposed in this paper has a large advantage in the recognition of named entities in Chinese electronic medical records,the fine-grained semantic representation learning algorithm of Chinese characters can deepen the semantic knowledge of Chinese electronic medical records,and the weak domain migration algorithm solves the problem of the scarcity of Chinese electronic medical records annotation corpus.