基于大数据架构的医疗档案分析平台的构建

The construction of analysis platform of medical archives based on big data architecture

目的:从当前医疗档案应用需求出发,拟构建适应现代医学发展的医疗档案分析平台,为医疗业务数据提供建档、检索查询及挖掘分析工具,并为医疗决策提供数据支持。方法:通过医疗档案管理及应用需求分析,结合流数据载入、大数据分布式存储、内存数据挖掘及数据可视化等关键技术设计,构建医疗档案分析平台。平台采用Hadoop分布式文件系统(HDFS)和分布式文件存储框架作为医疗档案大数据存储方案,分别在Spark SQL数据查询工具、Spark Mllib机器学习算法库及其挖掘分析流水线的基础上,开发实现数据整合、多维数据查询及分析挖掘组件,完成基于Echarts可视化组件的数据可视化。结果:构建的适应现代医学发展的医疗档案分析平台,能够帮助用户实现异构数据管理,实现海量数据查询及分析数据特征及规则;构建了基于知识库体系的应用组件,为医疗决策提供数据依据支持。结论:基于大数据架构的医疗档案分析平台能够提高医疗业务数据管理及分析的科学性和规范性,降低应用难度,可为其他同类应用系统构建提供参考。

Objective:To construct an analysis platform of medical archives that could adapt to the development of modern medicine based on the application requirement of medical archives at the present so as to provide the tools of establishing archives,searching and inquiring,and mining and analysis for medical business data,and to provide data support for medical decision-making.Methods:The analysis platform of medical archives was constructed by the management of medical archives and the analysis of application requirements,and by combining with the key techniques such as stream data loading,distributed storage of big data,memory data mining and data visualization and so on.The platform adopted HDFS distributed file storage frame as the storage scheme of big data of medical archives to develop and realize the components with data integration,multi-dimensional data query and analysis mining based on the Spark SQL data query tool,Spark Mllib machine learning algorithm bank and its assembly line of mining and analysis,respectively.Finally,the data visualization based on Echarts visualization component was completed.Results:The constructed analysis platform of medical archive,which adapted to the development of modern medicine,could help users to realize management of heterogeneous data,and realize massive data query and analyze data characteristics and rules.And it constructed application components based on knowledge bank system to provide data basis support for medical decisionmaking.Conclusion:The analysis platform of medical archives based on big data architecture can not only improve the scientificity and standardization of management and analysis of medical business data,and reduce the application difficulty,but also provide reference for the construction of other similar application systems.