Research And Realization Of Real - Time Distributed Retrieval Technology For CDA Clinical Documents

In recent years, massive number of Electronic Medical Records (EMRs) is produced in daily clinical activities which contains a lot of valuable information. How efficient use of clinical data for clinical documentation, research and education to provide services is becoming increasingly important, it depends on the efficient and accurate search technology to support.Efficient and precisely retrieval of relevant EMRs from a large-scale EMRs is becoming a big challenge, which includes three main aspects:(1) Clinical Document has a huge amount of data and distributed widely.So it be leading to clinical data too difficult to obtain real-time document indexing and retrieval;(2).It necessary to further improve the medical terminology complexity and ambiguity result retrieval precision.(3). Clinical data file format type and representation are complex,and it difficult conducive to use to achieve personalized display. Efficient use of domestic and foreign scholars have done a lot of research work on electronic medical records, effectively increasing the use value of clinical documentation data. However, The query efficiencies based on centralized methods are not suitable for handing medical big data because the response time is linearly increasing with the size of the datasets or concurrent end users. Besides, it has been used in clinical document retrieval based on semantic query expansion, but due to the complexity of TCM (Traditional Chinese medicine) Terms and fuzziness of retrieval accuracy are still needs to be improved. Finally, most of the existing research for a particular data format, It’s difficult to form a common document retrieval models and clinical reports personalized display and increased complexity of the clinical document retrieval system’s development.Regarding the issue above,we propose an efficient and robust framework for implementing a large-scale EMRs retrieval system in cloud environment (C-MRRS, Cloud-Medical Records Retrieval System). We first implement a parallel method based on MapReduce to improve the performance of index building and a distributed search cluster to provide high concurrent online EMRs retrieval. A multi-indexing model is proposed to ensure the latest EMRs are indexed and retrieved in real-time, and semantics-based query expansion and multi-factor ranking model are proposed to improve retrieval quality. To help users view the interested EMR reports, we implemented template-based visualization method for displaying the contents of medical reports. Furthermore, to better describe how C-MRRS provides retrieval service for users, a running example for diagnosis based on C-MRRS is discussed. The results show that the latest EMRs are indexed in real-time, the high relevant EMRs are retrieved efficiently from large-scale EMRs and the contents of medical reports are displayed via a friendly web page. Furthermore, the proposed system is more easily integrated with existing clinical systems and it can be used in various scenarios such as diagnosis, research and education.