Using automated speech processing to improve identification of risk for hospitalizations and emergency department visits in home healthcare

项目总结 每年，全美约有11000家家庭医疗保健(HHC)机构为更多的人提供护理 超过500万老年人。目前，大约三分之一的HHC患者正在住院或就诊。 如果给予适当和及时的护理，这些事件中高达40%是可以预防的。然而， 这些数字在过去十年中并没有改善，尽管国家和地方都在努力提高质量。 数据科学的一个子领域--自动语音处理--的最新进展开启了一种 未开发的丰富数据流，可通过分析护士与患者之间的语言来改进风险识别 沟通。这项拟议的研究汇集了家庭保健领域的一个跨学科专家团队 护理、自动语音处理、自然语言处理和风险模型开发的探索 自动语音处理能否提高家庭医疗中高危患者的及时识别能力 并可能减少他们的住院和急诊室就诊。 具体地说，本研究的目的是：目标1：完善并最终确定一个自动语音处理系统 确定患者与护士口头交流中住院和急诊科就诊的危险因素。目标2：探索 从病人-护士交流中提取的数据在多大程度上可以改善住院风险预测 与仅基于电子健康记录数据的风险模型进行比较时，ED就诊。 这项研究将建立首个住院和急诊风险模型，该模型自动合并 病人与护士言语交流的数据。在今后的工作中，该风险模型可以集成到家庭中 医疗保健临床工作流程，以触发有关患者趋势的及时和个性化警报， 进而启动适当和及时的护理，以防止HHC可避免的住院和急诊室就诊。

PROJECT SUMMARY Every year, about 11,000 home healthcare (HHC) agencies across the United States provide care to more than 5 million older adults. Currently, about one in three HHC patients are hospitalized or visit an emergency department (ED) and up to 40% of these events are preventable with appropriate and timely care. However, these numbers have not improved over the last decade, despite national and local quality improvement efforts. Recent advances in a subfield of data science—automated speech processing—have unlocked an untapped rich data stream that can improve risk identification by analyzing nurse-patient verbal communication. The proposed study brings together an interdisciplinary team of experts in home healthcare nursing, automated speech processing, natural language processing, and risk model development to explore whether automated speech processing can improve timely identification of patients at risk in home healthcare and potentially reduce their hospitalizations and ED visits. Specifically, the aims of this study are: Aim 1: Refine and finalize an automated speech processing system to identify hospitalization and ED visit risk factors in patient-nurse verbal communications. Aim 2: Explore to what extent data extracted from patient-nurse communications can improve risk prediction for hospitalizations and ED visits, when compared against the risk model based on electronic health record data only. This study will build a first-of-a-kind hospitalization and ED visit risk model that automatically incorporates data from patient-nurse verbal communication. In future work, this risk model can be integrated into home healthcare clinical workflows to trigger timely and personalized alerts about concerning patient trends, which will in turn activate appropriate and timely care to prevent avoidable hospitalizations and ED visits from HHC.