Structural Analysis And Research Of Electronic Medical Record Data

With the development of information technology,the data analysis and research results are gradually applied to various fields of social development.In the field of medicine,the structural analysis and research of electronic medical record data have gradually been favored by researchers and become the first work in the field of medical research.EMR(Electronic medical records,EMR)produced in the process of clinical treatment,contains all personal health information and consultation,and contains a large number of closely related to personal health medical knowledge,for named entity recognition and extraction of electronic medical records is an important data to study medicine.In electronic medical records,unstructured text data is the majority.However,unstructured data cannot be automatically analyzed and processed by computer,which hinders the mining and discovery of medical knowledge to a certain extent.This paper mainly carries out the work in the following aspects:(1)Through reading and studying a large number of electronic medical record data,the customized medical dictionary is obtained through analysis and collation.Meanwhile,the N-shortest path word segmentation method is improved through dynamic deletion algorithm.By optimizing the dictionary and method of Chinese word segmentation,the speed and quality of word segmentation are considered.(2)According to the text characteristics of the electronic medical record,this paper designed a kind of hidden markov model based on double tagging named entity recognition method of electronic medical records,the medical record text tagging part of speech and entity type at the same time,the optimization of hidden markov model decoding problem,so as to improve the accuracy of electronic medical records in the text entity recognition and recall rate.(3)According to the text characteristics of electronic medical records,this paper proposes a hidden markov model based on double tagging named entity recognition method of electronic medical records,as well as the basic parts of speech and the entity type,medical record label text optimized decoding problem for hidden markov models,thus improve the named entity recognition in the text of the electronic medical record accuracy.(4)In order to meet the requirements of upper application service research on data types,this paper normalizes the extraction results of named entities and their modified values to form structured data of different data types.(5)Based on the technical route of the study and the experimental results in the research process,this paper designs and develops a structured analysis system of electronic medical record data.The system integrates the improved algorithm and model in this paper,and presents the data experimental test results.This paper takes the medical record text of physical examination as an example to conduct structural analysis and research on electronic medical record data.Using electronic medical records data research method proposed in this paper,high quality to identify and extract named entities in the medical records and the corresponding modify values,and the transformation from unstructured data to structured data is realized.By normalizing the extraction results,structured medical data support is provided for information retrieval,data mining and clinical decision support in the medical field.