支持区域医疗和分级诊疗的数据互操作与隐私保护数学技术

Advancing studies on mathematical algorithms and key technologies for medical data interoperability and privacy protection can help solve an array of bottleneck problems such as data sharing, transmission, privacy protection and confirmation of data rights, which is a fundamental and strategic effort for developing intelligent and inter-connected hospitals in China. This study plans to innovatively explore intelligent semantics understanding and inference technologies driven both by data records and ontologies, aiming to eliminate structural and semantic heterogeneity among multiple databases such that concepts can be aligned and content can be bi-directionally translated among multiple heterogeneous databases with a high performance. We plan to explore the new idea of incorporating blockchain technologies with virtual integration technologies to enable various medical data interoperability operations such as data acquisition and data right confirmation across multiple systems. We will also explore a series of privacy protection technologies, such as sandbox and homomorphic encryption techniques customized to satisfy application needs raised by various medical scenarios and tasks. In recognition of Chinese own national conditions, the project further proposes a novel data exchange method based on the notion of data resource objects along with a set of standardized and general-purpose interfaces for data storage, exchange, privacy protection and confirmation of data rights to be used by tiered intelligent and interconnected hospitals in China. These interfaces support data sharing among multiple institutes with a high efficacy and level of security. By integrating all algorithms developed in the above, the project will build a full-fledged data management and application platform. Last but not least, the project will verify the feasibility and adoptability of the proposed technical system through case applications such as similar medical case retrieval across institutions and construction and utilization of patient profiles across regional healthcare systems. Overall, the execution of this project can greatly facilitate medical data sharing with a high efficacy, which in turn can effectively help advance the implementation of regional and tiered medical services.

开展医疗数据互操作与隐私保护相关数学算法和关键技术研究，解决数据共享、流通、隐私保护和确权等瓶颈问题，是推进我国智联网医院发展的基础性和战略性工作。本项目拟创新性地探索数据与本体双驱动的智能语义理解和推断技术，消除数据库间的模式和语义异构，实现多源异质多数据库间高性能地概念对齐与内容互译；并探索区块链与虚拟集成技术融合的新思路，实现跨系统的医疗数据获取、确权等数据互操作功能；针对不同场景任务需求，探索沙箱、同态加密等多重隐私保护技术；进一步提出了一种基于数据资源对象的交换方法以及一套面向我国国情的通用智联网医院分级数据存储、交换、隐私保护与确权标准接口，支持多机构间高效、安全的数据共享；集成上述算法，研发一套完整的数据管理与应用平台；最后，通过跨机构相似病历查询和区域性患者画像等场景验证所开发技术体系的可行性和可推广性。项目的实施将有助于医疗数据流通并促进区域医疗和分级诊疗。

开展医疗数据互操作与隐私保护相关数学算法和关键技术研究，解决数据共享、流通、隐私保护和确权等瓶颈问题，是推进我国智联网医院发展的基础性和战略性工作。本项目创新性地探索了数据与本体双驱动的智能语义理解和推断技术，旨在消除数据库间的模式和语义异构；在此基础上提出了一种多源异质多数据库间高性能地概念对齐与内容互译的算法解决方案，并开发了相应的软件原型验证系统。项目研究同时也探索了区块链与虚拟集成技术融合的新思路，并以软件原型系统的形式实现与验证了跨系统间医疗数据获取、确权等数据互操作功能的可行性及性能优势。本研究还针对不同场景任务需求，探索了沙箱、同态加密等多重隐私保护技术，开发了相应的软件原型验证系统。在此基础上，项目进一步提出了一种基于数据资源对象的交换方法以及一套面向我国国情的通用智联网医院分级数据存储、交换、隐私保护与确权标准接口，用以支持多机构间高效、安全的数据共享，并逐一开发了相应的软件原型验证系统。项目在集成上述算法设计与软件原型系统实现的基础上，进而研发了一套完整的数据管理与应用平台，可支持智联网医院开展一系列面向单体医院的“互联网+健康”综合医护服务以及面向医联体的协同诊疗及健康管理服务。最后，基于上述数据共享技术获得的跨机构和系统的多源异质医疗数据，本项目进一步开展了一系列医疗数据智能分析、模型设计、训练与优化的工作，包括面向肺部和皮肤疾病的医学图像分割、病灶分类、辅助诊断与智慧护理支持算法、模型与系统的设计和研发。上述“互联网+健康”示范应用和计算机辅助临床决策支持模型与系统的成功研发实证了在实现医疗大数据安全共享及流通后，开展线上、线下一体化“互联网+智慧医护服务“及人工智能医疗辅助决策模型的可行性与性能优势。项目的实施有助于借助人工智能和互联互通医疗大数据的技术驱动力，促进区域医疗和分级诊疗的深化落地。