Research And Implementation Of Structuring Medical Record Of Thyroid Disease Based On Deep Learning

With the development of informatization in major hospitals in China,various clinical information systems have accumulated rich clinical data resources for hospitals,including images and texts.However,medical data is growing geometrically,and it is extremely difficult to extract key information from texts by traditional methods,and a large amount of unstructured data also becomes an obstacle to information sharing among hospitals.In conclusion,the traditional method based on manual extraction of key information has been unable to adapt to the study of massive data and unstructured data in the era of big data.Therefore,the research on structured processing of unstructured medical text data is of great significance.This paper focuses on the medical record of thyroid disease in clinical text data,and carries out research work based on the medical record of thyroid disease in the third-grade class-A hospital in Shanghai,and implements a complete set of structured processing algorithms based on deep learning.The main contributions of this paper are as follows.1)In the process of data preprocessing,this paper proposes an iterative professional lexicon construction method for the problem that the medical knowledge of thyroid disease is involved in a large number of medical professional knowledge,which can be used to guide the word segmentation and improve the accuracy of word segmentation.2)In the process of structured processing,aiming at the problem that the traditional dictionarybased or semantic-based information extraction methods cannot give consideration to both generalization and accuracy,this paper proposes an integrated information extraction method combining entity recognition and thesaurus matching algorithm,using different information extraction algorithms for different medical record contents.3)In the structured storage process,aiming at the problem that the traditional structured representation method cannot provide convenient support for the storage,analysis and retrieval of knowledge,this paper summarizes the hierarchical structure and characteristics of the text data of the medical record of thyroid disease based on the existing data,and designs the knowledge ontology in this field.4)In the realization process of medical record of thyroid disease structured system,this paper analyzes the actual needs of users,designs the system framework based on demand analysis,and presents and explains the interface of each functional module of the system in combination with the specific implementation mode of the system.5)In the process of analyzing the experimental results,precision rate,recall rate R and F score are used to demonstrate the experimental results.And,this paper designs a comparative experiment to verify the necessity and effectiveness of the structured method proposed in this paper.In summary,this paper uses the medical record of thyroid disease of the third-grade class-A hospital in Shanghai as experimental data,and propose a structuring method based on deep learning,which can structure the unstructured medical record of thyroid disease and save the results in RDF.Experiments show that this method can achieve the expected goals and provide data support for subsequent research of medical big data.Based on the implemented algorithm,this paper designs and implements a medical record-structured system of thyroid disease,and the effectiveness of the structured method proposed in this paper is verified through comparative experiments.