Advancing Interdisciplinary Science of Aging through Identification of Iatrogenic Complications: The UF EHR Clinical Data Infrastructure for Enhanced Patient Safety among the Elderly (UF-ECLIPSE)

项目概要/摘要 医源性疾病是一个持续的公共卫生问题，据估计， 每年有5万名老年人在美国医院接受治疗。医院获得性福尔斯和医院- 诱导性谵妄是最常见和最昂贵的医源性疾病，其发生率是 彼此相连计算技术的进步和电子数据的可用性 有机会更准确地识别有医院获得性跌倒风险的患者，或 医院诱发的精神错乱目前，美国约80%的医疗机构正在通过电子方式获取临床数据。 人口大约75-80%的临床数据是文本数据，无法使用传统的 统计方法开发研究数据基础设施，支持使用文本和 结构化数据对于旨在改善护理和患者结果的学习型卫生系统至关重要。 在这个项目中，我们建议扩大电子数据驱动知识的研究基础设施 通过佛罗里达大学（UF）的EHR数据基础设施的发展， 老年人安全（UF-ECLIPSE）。我们研究计划的长期目标是提高 通过有效的学习健康，减少医源性疾病，提高住院老年人的安全性 系统具体目标1（R21阶段）：确定并测试可行性 文本挖掘管道处理注册护士（RN）的进度记录，以预测医院获得性 福尔斯。我们将采用有监督和无监督的文本挖掘方法来识别文本 与患者福尔斯相关的属性。然后，我们将利用患者跌倒风险因素的预测模型 在以前的工作中开发，以生成文本和结构化数据的复合模型，以预测 病人摔倒具体目标2（R33阶段）：确定和评估一个组织的结构和人力资源， 扩大研究数据基础设施，以支持持续的跨学科老龄化研究。创新和 试点测试文本挖掘管道，以生成医院引起的谵妄的预测模型。然后我们将 将开发的管道集成到现有的UF健康临床数据仓库（CDW）基础设施中 并进行测试以评估功能性、耐用性和可扩展性。此外，我们还建议开发人类 资源基础设施，以支持数据驱动的跨学科老龄化研究。这将通过培训来实现 研究生在跨学科的数据科学为老龄化研究。 UF-ECLIPSE研究团队将成为首批实施和测试集成数据存储库的团队之一 它利用护士生成的结构化和文本数据来支持学习健康系统。这项研究将创造 重要的新研究数据基础设施，并将成为医疗保健组织提高安全性的典范。 为每天住院的数百万美国老年人提供有效的护理。

Project Summary/Abstract Iatrogenic conditions are a continuing public health concern, causing death among an estimated two hundred and fifty thousand older adults annually in United States (US) hospitals. Hospital-acquired falls and hospital- induced delirium are among the most common and costly iatrogenic conditions, and their occurrences are linked to each other. Advances in computing technology and availability of electronic data presents opportunities to more accurately identify identifying patients at risk of suffering a hospital-acquired fall or hospital-induced delirium. Clinical data is now being captured electronically for about 80% of the US population. Approximately 75-80% of clinical data is text data which cannot be analyzed using traditional statistical methods. The development of a research data infrastructure that supports the use of text and structured data is critical for a learning health system aimed at improving care and patient outcomes. In this project, we propose to expand the research infrastructure for electronic data-driven knowledge generation through the development of the University of Florida (UF) EHR Data Infrastructure for Patient Safety among the Elderly (UF-ECLIPSE). The long-term goal of our research program is to enhance the safety of hospitalized older adults by reducing iatrogenic conditions through an effective learning health system. We plan to carry out the following aims: Specific Aim 1 (R21 Phase): Identify and test the feasibility of text-mining pipelines to process registered nurses' (RNs) progress notes for prediction of hospital-acquired falls. We will employ a combination of supervised and unsupervised text-mining methods to identify text attributes associated with patient falls. We will then leverage a predictive model of patient fall risk factors developed in previous work to generate a composite model of text and structured data to predict the odds of a patient falling. Specific Aim 2 (R33 Phase): Determine and evaluate the structural and human resources of an expanded research-data infrastructure to support sustained interdisciplinary aging studies. We will develop and pilot test text-mining pipelines to generate a prediction model of hospital-induced delirium. We will then integrate the developed pipelines into the existing UF Health Clinical Data Warehouse (CDW) infrastructure and test to assess functionality, durability and scalability. In addition, we propose to develop the human resource infrastructure to support data-driven interdisciplinary aging research. This will be achieved by training graduate students in interdisciplinary data science for aging research. The UF-ECLIPSE research team will be among the first to implement and test an integrated data repository that utilizes nurse-generated structured and text data to support a learning health system. This study will create important new research data infrastructure, and will be a model for health care organizations to increase safe effective care for the millions of older adult Americans hospitalized every day.