Design And Implementation Of An EMR Anonvmization System Based On NER

As with the widespread adoption of information technology in hospital modernization construction, intellectual management and recording of the entire medical treatment of patients has been realized through electronic medical record (EMR), which becomes the major medium carrying medical data with a dramatic geometrically increase. For this reason, it has been a hot spot of study on exploiting valuable medical knowledge from the massive data recorded in EMR. However, medical data has to be anonymized prior to being extracted, to protect patients’ private information. Yet how to keep the data complete while trying to prevent privacy from leakage turns to be a great challenge. This paper presents an anonymization system of electronic medical record documents based on NER to meet this challenge.This paper is basically structured as three sections:CA identity verification security service, protected health information identification module, and encryption transmission service of medical documents. CA identity verification is able to efficiently safeguard the web security and prevent web attack and data plagiarism from illegal users. Among the identification of protected health information, the general recall and accuracy of private information identification both exceed 90%, far above individual manual inspection at 81% averagely and is close to manual inspection by two persons at 94% averagely. Documents encryption transmission fills the gap of FTP in plaintext data transmission security and guarantees data confidentiality as well. The anonymization system of EMR documents described in this paper has eventually fulfilled the purpose of patient privacy protection with the integrated use of three major modules.