节点文献
临床过程分析与优化技术研究
Research on Clinical Process Analyzing and Opti-mizing Technologies
【作者】 黄正行;
【导师】 段会龙;
【作者基本信息】 浙江大学 , 生物医学工程, 2010, 博士
【摘要】 随着医药卫生事业的发展,临床过程越来越专业化和精细化。临床过程分析与优化着重在辨识出一连串的临床诊疗活动,通过科学分析方法对这些诊疗活动的流程进行分析与优化。其目标是确保临床诊疗活动的执行能具有一定的水准和精确度,并能得到持续性的改善,使得临床诊疗服务水平不断提高。临床过程分析与优化技术研究是近年来国内外医学信息学的研究热点。已有的研究成果在应用时存在较大的局限性,主要原因是难以满足临床过程的复杂性和灵活性要求,且缺乏能根据医学知识发展和临床环境变化而持续优化与完善的科学分析方法。为此,本论文通过分析临床过程实践的主要特点,将临床过程实践视为一个包括过程表达、执行、分析和优化等多个阶段的完整生命周期的行为,研究了该生命周期中过程表达、过程推荐、知识检索、资源分配、过程评估和分析等关键技术问题,以此为基础初步建立了临床过程分析与优化技术研究体系,主要包括:●建立符合患者临床诊疗实践的过程表达方法是临床过程分析与优化的基础。针对临床过程灵活多变、变异频繁发生的特点,本论文提出一种分层次的临床过程表达方法,研究临床诊疗过程不同层次的过程表达模型及实现技术,支持临床过程内容和过程结构的动态调整,提高了临床过程表达的灵活性。●临床过程执行控制是实现临床过程可靠、有效运行的保证。本论文根据临床环境变化对过程执行控制影响较大的特点,建立了自适应的临床过程执行机制,能够在临床过程实践中,根据患者临床状态的变化,选择和推荐适合患者当前状态的临床诊疗计划,依据循证医学原则分析和保证患者的最佳诊疗预后。●临床过程注重知识使用的准确性和时效性。这要求能够在临床过程中,可以及时为医护人员的诊断治疗工作提供有效的医学知识。现有的知识检索方法或是不能够满足临床过程实践对知识检索的时间要求,或是检出结果准确率较低,难以满足医护人员的实际需要。本论文提出一种医学知识检索和推荐方法。该方法融合了关键词检索技术和协同过滤技术,能够提高医护人员临床工作中的医学知识检索的时效性,促进临床各专科医护人员之间的协作。·资源合理分配是临床过程高效执行的重要前提。现有的临床过程资源分配机制比较简单,大多关注于资源的权限管理,很少考虑资源分配对临床过程执行性能的影响。本论文提出了一种基于强化学习的临床过程资源分配方法。在临床过程实践中,该方法通过与临床过程环境的实时交互,学习和推理适合当前临床环境的资源分配最佳策略,并在之后相同或相似的临床环境中使用,提高资源的使用效率,改善临床过程的执行性能。●临床过程执行时,与诊断治疗行为有关的临床信息和过程信息会被各类医疗信息系统记录并保存在临床数据库中,这些信息表达了患者就诊期间的所有行为记录,能反映临床过程执行时的真实情况,为临床过程的分析和优化提供了必要的数据依据。为次,本论文提出基于过程挖掘技术的临床过程评估和分析方法,通过收集、挖掘和分析临床过程执行信息,构建临床过程模型,开展临床过程执行的一致性检验,为临床过程的评估、分析和优化提供技术支持。本论文在临床特定专科中有针对性的临床过程表达、过程执行、知识推荐、资源分配、过程挖掘等关键技术开展临床实践评估。结果表明,本论文的研究成果能够合理的表达临床过程模型,有效的推荐适合患者状态的医疗干预,灵活的处理过程变异,分配合适的资源开展临床诊疗活动,及时的为医护人员提供所需要的医学知识,促进医护人员之间的协作,优化临床诊疗活动的执行效率,帮助医护人员分析患者临床诊疗行为,改进与优化临床过程。通过融合以上研究成果,本论文研究和实现了一系列临床过程分析与优化技术,通过与其它临床信息系统和功能模块集成,形成面向大中型医院的完整的临床过程分析与优化信息系统框架。本论文是针对临床过程分析与优化技术研究的有益探索和尝试,为最终形成高效的面向临床过程的医疗信息系统平台奠定良好的技术基础。
【Abstract】 Along with the development of medicine and technology, clinical process has become more and more specific and subtle. Clinical process analysis and optimization can help sort out the concept of medical practice that could be covered by scientific approaches, improve patient care services, and bring the best clinical prognosis through the scientific methods.The research on clinical process analyzing and optimizing technologies is becoming an important focus in clinical engineering and medical informatics domain. However, the characters, such as complexity and flexibility of clinical processes, result in the absence of real progress towards applying advances in information technology in order to analyze and optimize clinical processes. Especially, it lacks the scientific methods aimed at continuous optimization of clinical processes within clinical environmental changes. Considering the challenges above, this thesis introduces a detailed analysis of clinical processes, presents an in-depth study on the key technologies of clinical process analysis and optimization, i.e., process representation, execution control, medical knowledge retrieval and recommenda-tion, resource allocation and process analysis, and builds an initial technology framework of clinical processes analysis and optimization. The main work includes the following:An appropriate clinical process model is the key issue of the efficiency of clinical process practice. This thesis presents a hierarchical clinical process modeling mech-anism. It exploits the later binding modeling technique to support the content changes and the structure adjustments of clinical processes, and enhances the clinical process modeling flexibility.In order to guarantee the reliable clinical process execution, this thesis presents an adaptive clinical process execution mechanism, which can recommend appropriate medical interventions within the changes of patient state in clinical processes.Clinical practice heavily depends on the efficiency of knowledge retrieval and uti- lization. It requires to provide medical knowledge timely and accurately in clinical processes. This thesis proposes a clinical process oriented medical knowledge rec-ommendation mechanism, which can improve the efficiency of medical knowledge retrieval and assist medical staffs’work in clinical processes. The proposed solu-tion is easy to implement and can benefit medical staffs by facilitating collaboration work.Resource allocation is an integral part of clinical processes. It is widely acknowl-edged as being important for the efficiency of clinical processes. This thesis intro-duces a reinforcement learning based clinical resource allocation mechanism, which can optimize resource allocation and improve clinical process performance.In clinical practice, information about patient care are regularly recorded in vari-ous medical information systems and stored in clinical databases. Those information express the medical behaviors of clinical processes. This thesis proposes a process mining based clinical process analysis and evaluation mechanism in order to sup-port clinical process (re)design and optimization through the collection, mining and analysis of clinical process execution information.The efficiency and usability of the proposed approaches is validated by a set of experi-ments in specific clinical departments. The experimental results indicate that the proposed approaches can effectively solve the key challenges of clinical processes, optimize the per-formance of the executions of clinical processes, and improve the quality of patient care. By fusing the proposed research results, a clinical processes analyzing and optimizing tech-nology framework is designed and implemented. The methods and practice can favor the dissemination of clinical process analysis and optimization, and can be beneficial for med-ical staffs, patients and hospital managers.