Research Of Large-Scale Electronic Medical Record Text Analysis System Based On Cloud Computing

Electronic Medical Record (EMR) is an important research field of health and medical informatization. As complete and detailed clinical information resources of being produced and recorded during patients'treatment, structured electronic medical record also contains a large number of unstructured text information, such as medical records of clinical manifestations recorded by natural language. Electronic medical record data is large in a hospital or cross hospitals area, how to annotate and analyze unstructured text information within large-scale electronic medical record data and build index for query is an urgent problem to be solved.For the problem above, this paper presents a solution of analyzing the unstructured text information of a large volume of electronic medical records and building index based on UIMA in cloud computing environment, designs and implements a prototype system after deeply researching unstructured information management architecture UIMA (Unstructured Information Management Architecture) specification and cloud computing programming model MapReduce and other key technologies.Contrasted with the traditional text analysis systems, this paper has the following characteristics:1) Combine UIMA framework with cloud computing programming model MapReduce, a solution of analyzing the unstructured text information of a large volume of electronic medical records and building index based on UIMA in cloud computing environment is proposed. This solution not only take advantage of parallel processing capacity of cloud computing environment based on MapReduce, but also keep the openness of UIMA framework which can develop and deploy different Analysis Engine according to different requirement.2) The prototype system based on the solution above supplies the interface of electronic medical record repository based on XDS (Cross-Enterprise Document Sharing) and preprocesses the unstructured text information of electronic medical records according to the input requirements of cloud computing platform Hadoop. The prototype system can analyze the unstructured text information and build index parallelly.3) Realizing a Chinese Analysis Engine based on UIMA specification. This Analysis Engine is based on open source Chinese word segmentation software IKAnalyzer using external CMV (Controlled Medical Vocabulary) and can annotate unstructured Chinese text information recorded by natural language of structured electronic records.Experimental data and the effect of the prototype system show that the system is feasible and effective.