MULTIMEDIA ARCHITECTURE FOR THORACIC DISEASES

医生在从现有的医疗数据输入系统中检索重要的患者信息时面临着很大的困难。这些困难分散了临床医生对患者直接护理的时间，有时患者的基本信息仍未被发现。以概念为中心，以时间为导向的医疗数据组织可以提高医生从病历中收集重要数据的效率。数据检索效率的提高很可能会对护理过程、医生满意度和患者满意度产生积极影响。为了验证这些假设，本研究将扩展之前在肿瘤学多媒体数据库上的工作，创建一个名为TimeLine的软件系统，该系统使用特定领域的本体和可调节的时间粒度来呈现临床数据。时间表将首先使用传统的信息检索措施进行技术评估。此后，TimeLine将采用时间序列设计在两个临床环境中部署。该项目的技术目标将涉及肺病领域的概念表征和数据检索。将开发一个特定领域的知识库，根据器官系统、医疗状况、症状或其他概念的高级抽象对数据进行建模。计算机界面将为不同的医生档案设计，但将能够适应时间抽象和细节深度的不同要求，这取决于医生的信息需求。该系统的实现将采用万维网技术。系统评估将分三个部分进行。自动数据预取的准确性以及在抽象（器官系统，问题等）的增量级别上的数据组织将使用精度和召回的经典度量来评估。在对界面进行最终改进后，将使用时间序列“开-关”试验设计，将TimeLine对护理过程和医生满意度的影响与真实临床环境中的现有数据信息环境进行比较，在该试验设计中，肺科医生在响应住院患者的咨询过程中使用传统的TimeLine，然后使用传统的数据环境。使用来自账单记录、医疗记录审查和时间运动研究的信息，每个病人的临床稳定时间和资源利用将作为过程变量进行测量。医师满意度将使用标准化仪器进行评估。最后，在加州大学洛杉矶分校胸部肿瘤诊所使用类似的时间序列试验设计，测量肿瘤放射科医生和肿瘤科医生对数据环境的满意度。还将在各自的数据环境中探讨患者生活质量措施。这将是第一个将详细的概念模型和接口与基本信息检索措施联系起来的研究，并确定在医疗保健过程中改进信息环境的价值。

Physicians face significant difficulties retrieving essential patient information from existing medical data entry systems. These difficulties divert clinicians' time away from direct patient care, with essential patient information sometimes remaining undiscovered. Concept-centered, time- oriented medical data organization should improve physicians' efficiency in gathering essential data from medical records. It is likely that improved efficiency in data retrieval should favorably affect processes of care, physician satisfaction, and patient satisfaction. To test these hypotheses, this study will extend previous work on oncology multi-media databases to create a software system, called TimeLine, which presents clinical data using a domain-specific ontology and an adjustable temporal granularity. TimeLine will undergo technical assessments first using traditional information retrieval measures. Thereafter, TimeLine will be deployed in two clinical settings using time-series designs. The technical objectives of the project will address concept characterization and data retrieval in the domain of pulmonary diseases. A domain-specific knowledge base will be developed that models data according to high level abstractions of organ system, medical condition, symptom, or other concept. The computer interface will be designed for various physician profiles, but will be capable of adapting to varying requirements of temporal abstraction and depth of detail, depending upon the information needs of the physician. Implementation of the system will use World Wide Web technology. System evaluation will proceed in three parts. The accuracy of automated data pre-fetching as well as data organization at incremental levels of abstraction (organ system, problem, etc.) will be assessed using the classical measures of precision and recall. After final refinements have been made to the interface, the influence of TimeLine on processes of care and physician satisfaction will be compared to existing data information environments in real clinical settings using a time-series "off- on-off" trial design in which pulmonologists use traditional, than TimeLine, then traditional data environments in the course of responding to consultations on hospitalized patients. Using information from billing records, medical records reviews, and time-motion studies, each patient's time to clinical stabilization, and resource utilization will be measured as process variables. Physician satisfaction will be assessed using standardized instruments. Finally, using a similar time-series trial design in UCLA Thoracic Oncology Clinic, oncologic radiologists' and oncologists' satisfaction with the data environment will be measured. Patient quality of life measures will also be explored in the respective data environments. This will be the first study to link a detailed conceptual model and interface with measures of essential information retrieval and to determine the value of an improved information environment in healthcare processes.