医疗大数据平台建设需求、实施路径与成效探讨

Research on demand, implementation pathways and effectiveness of medical big data
platform

  • 摘要:
    背景 随着医疗信息化的发展,医疗大数据平台成为临床研究资源再分析、利用的关键突破点。然而,医疗数据 的多源异构性、数据标准不一致、患者隐私数据安全等特点增加了数据采集与应用难度。目的 分析医疗大数据平台建设 需求,研发自助式全流程数据治理平台及工具,构建多中心医疗大数据平台。方法 随着大数据、人工智能等技术的发展, 通过梳理医院数据应用需求,采用模块化、组件化的构建思路进行平台架构设计,提炼出与应用系统相对独立、通用的组 件及管理工具,搭建多中心、多源异构医疗大数据平台。结果 完成解放军总医院门急诊和住院电子病历数据汇聚治理, 研发了全流程、可视化数据治理工具。定义的事件图谱schema涵盖29个本体类别和128个概念、1 009种关系和3 022种属 性,涵盖临床循证医学知识和临床诊疗、物联网、医学影像等数据。数据治理的一致性和可溯源性达99. 99%,知识准确率 达到95%以上。构建了贯穿科研全流程,覆盖不同研究类型需求的一站式数据智能检索与科研分析系统和专病库智能分析 系统。结论 该平台不仅面向临床科研人员提供了数据检索与分析能力,还为数据工程师提供了数据治理和平台运维工具, 提升了平台的可扩展性和灵活性。

     

    Abstract:
    Background With the development of medical informatization, medical big data platforms have become an important foundation for clinical research and a key breakthrough point for resource re analysis and utilization. However, the multi source heterogeneity of medical data, inconsistent data standards, and patient privacy data security have increased the difficulty of data collection and application.Objective To analyze the requirements for establishing a medical big data platform, develop a self service, full-process data governance platform and tools, and construct a multi-center medical big data platform.Methods With the development of technologies such as big data and artificial intelligence, we have conducted a comprehensive review of hospital data application needs. By adopting a modular and component-based design approach, we have designed the platform architecture. We have extracted universally applicable components and management tools that are relatively independent of specific application systems. These components and tools have been utilized to build a multi-center, multi-source heterogeneous medical big data platform. Results The electronic medical record data of the outpatient, emergency and inpatient departments of the General Hospital of the People's Liberation Army have been aggregated and processed. A full-process, visual data governance tool has been developed. The defined event schema graph covers 29 ontology categories, 128 concepts, 1 009 relationships and 3 022 attributes, including clinical evidence-based medical knowledge, clinical diagnosis and treatment, Internet of Things, medical imaging and other data. The consistency and traceability of data governance have reached 99.99%, and the knowledge accuracy rate is over 95%. A one-stop data intelligent retrieval and scientific research analysis system and a specialized disease database intelligent analysis system covering the entire scientific research process and different research types have been constructed.Conclusion The platform not only provides data retrieval and analysis capabilities for clinical researchers, but also provides data governance and platform maintenance tools for data engineers, which improves the scalability and flexibility of the platform.

     

     

/

返回文章
返回