Meeting Clinicians' Information Needs with Highly Tailored Knowledge Summaries

描述(由申请者提供)：该项目专注于设计和评估方法，通过生成临床医生量身定做和患者特定的知识摘要来改进临床医生的决策。知识摘要将由与临床医生的患者特定信息需求相关的语义片段(即，提供有意义的信息的小文本单元)组成。这种决策支持很重要，因为临床医生经常在病人护理过程中提出信息需求，而这些需求在很大程度上没有得到满足。未得到满足的信息需求错失了自我指导学习和改善患者护理的机会。 虽然临床医生问题的答案经常可以在在线健康知识资源中找到，但显著的障碍限制了这些资源用于患者护理。一种日益流行的降低这些障碍的方法是在电子健康记录(EHR)系统中提供上下文相关的“信息按钮”链接。根据临床情况，信息按钮预测临床医生的信息需求，并提供相关的知识资源链接。信息按钮可以很好地帮助临床医生满足简单的信息需求，但当(I)在没有大量认知努力扫描检索到的信息的情况下无法轻松找到答案；以及(Ii)信息需求与电子病历屏幕上未显示的数据相关联时，信息按钮不是最优的。 在拟议的研究中，我们将解决以前利用重大初步研究、最先进的信息提取和文本摘要工具以及越来越多地采用电子健康记录标准的方法的局限性。这项研究将以信息寻求行为理论的基础为指导。这项研究的目的和假设如下： *利用患者的电子病历数据生成知识摘要。H1：在寻找患者特定问题的答案方面，知识摘要比手动搜索更有效和高效。 *确定有助于临床医生的信息需求和信息寻求行为的背景和认知因素。H1：(I)语境和认知因素与信息需求的类型有关；(Ii)与追求信息需求的决定有关。 *根据临床医生的背景和认知因素，将知识摘要转换为量身定制的知识摘要。H1：在满足临床医生的信息需求和改进决策方面，临床医生量身定制的知识摘要比知识摘要更有效和高效。

DESCRIPTION (provided by applicant): This project focuses on designing and evaluating methods to improve clinician decision-making by generating clinician-tailored and patient-specific knowledge summaries. Knowledge summaries will consist of semantic fragments (i.e., small units of text that provide meaningful information) that are relevant to a clinician's patient specific information needs. This kind of decision support is important because clinicians often raise information needs in the course of patient care and these needs are largely unmet. Unmet information needs are missed opportunities for self-directed learning and improved patient care. Although answers to clinicians' questions can often be found in online health knowledge resources, significant barriers limit the use of these resources for patient care. An increasingly popular approach to lowering these barriers is to provide context-sensitive "infobutton" links within electronic health record (EHR) systems. Based on the clinical context, infobuttons anticipate clinicians' information needs and provide relevant links to knowledge resources. Infobuttons do a good job helping clinicians' meet simple information needs, but are less optimal when (i) answers cannot be easily found without substantial cognitive effort scanning the information retrieved; and (ii) the information need is associated with data not displayed on the EHR screen. In the proposed study, we will address limitations of previous approaches leveraging significant preliminary research, state-of-the-art information extraction and text summarization tools, and increasingly adopted EHR standards. The research will be guided by a foundation of information-seeking behavior theories. The study has the following aims and hypotheses: * Generate knowledge summaries leveraging patients' EHR data. H1: Knowledge summaries are more efficacious and efficient than manual search for finding answers to patient-specific questions. * Identify contextual and cognitive factors that contribute to clinicians' information needs and information-seeking behavior. H1: (i) Contextual and cognitive factors are associated with the type of information need; and (ii) with the decision to pursue an information need. * Transform knowledge summaries into tailored knowledge summaries based on clinician's contextual and cognitive factors. H1: Clinician tailored knowledge summaries are more efficacious and efficient than knowledge summaries for meeting clinicians' information needs and improving decision-making.