FW-HTF-P: Understanding the Social and Ethical Implications of Algorithmic Decision Making by Nurses Using the Rothman Index in a Large-Scale Hospital Setting

随着向基于价值的医疗的转变，医疗保健组织正在寻找创新的方法来提高医疗质量，同时降低成本。在接下来的几十年里，算法驱动的患者护理和数据驱动的警告系统(如罗斯曼指数)的广泛使用将会增加。罗斯曼指数是一个算法系统，它使用患者信息、实验室结果和其他与健康有关的数据来预测健康恶化。罗斯曼等算法系统无疑将改变它对医疗保健领域日常工作的意义，要求本已过度扩张的劳动力提供新的专业知识和沟通。虽然这类系统可能会减少工作人员的需求，但它们并不能消除人为监督和干预的需要。因此，了解传统培训的医疗从业者如何与此类系统交互对于成功实施算法患者护理至关重要。为了实现这一目标，这个项目试图扩大我们对虚拟重症监护病房(VICU)中的护士如何使用罗斯曼指数做出算法知情的工作场所决策的理解。更广泛地说，该项目考察了医护人员如何在算法和患者互动之间进行调解。因此，它为未来的培训协议、系统重新设计和修订以及类似的未来算法工具的整体优化提供了信息。该项目将采用参与性和以工人为中心的设计流程，通过与医疗工作者的密切合作来实施算法患者护理，以推进罗斯曼指数的技术设计。它将侧重于虚拟护士的经验，即目前在罗斯曼指数警报和现场健康从业者之间进行调解的提供者在做出实际算法决策方面的经验。该项目的基础是与相关医院利益相关者建立关系，然后收集和分析有关在VICU工作环境中使用罗斯曼指数的初步数据，包括对虚拟护理人员的观察和采访，以及从罗斯曼指数系统提取的确定的定量使用数据。根据初步调查结果，医院利益相关者将参与设计，以改进围绕算法决策的工作场所协议和实践。随后，本项目将介绍三个基本研究挑战：(1)关于认知和记忆如何促进工作场所写作过程的技术传播研究，以及电子系统对患者-提供者和提供者-提供者沟通的影响；(2)关于成功实现VICU的协同机制的组织研究；(3)关于罗斯曼指数等算法干预的社会和伦理影响的计算机科学研究。此外，这项研究将为研究许多相关问题打开大门：(A)预防性人口外展计划的算法患者护理的可行性；(B)算法患者护理实施和预期劳动力培训导致的工作转移的预测；(C)医疗保健工作场所的未来算法设计。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

With a shift towards value-based care, healthcare organizations are seeking innovative ways to improve the quality of care while reducing costs. The next several decades will see a rise in algorithm-driven patient care and the widespread use of data-driven warning systems such as the Rothman Index. The Rothman Index is an algorithmic system that uses patient information, lab results, and other health-related data to predict health deterioration. Algorithmic systems such as the Rothman will undoubtedly change what it means to daily work in the field of healthcare, demanding new kinds of expertise and communication from an already over-extended workforce. While such systems may reduce staffing needs, they do not eliminate the need for human oversight and intervention. Understanding how traditionally trained healthcare practitioners interact with such systems is, therefore, vital for the successful implementation of algorithmic patient care. To accomplish this goal, this project seeks to expand our understanding of how nurses in the Virtual Intensive Care Unit (VICU) make algorithmically informed workplace decisions using the Rothman Index. More broadly, the project examines how healthcare workers mediate between algorithms and patient interaction. Consequently, it informs future training protocols, system redesign and revision, and overall optimization of similar futuristic algorithmic tools. This project will employ a participatory and worker-centered design process for implementing algorithmic patient care through close collaborations with healthcare workers to advance the technical design of the Rothman Index. It will focus on the experiences of the virtual nurses, the providers who currently mediate between the Rothman Index alerts and health practitioners on the floor, in making actual algorithmic decisions. The project’s foundation is relationship-building with relevant hospital stakeholders followed by collecting and analyzing preliminary data about experiences with the Rothman Index in the VICU work environment, including observations and interviews with virtual nursing staff and deidentified quantitative usage data extracted from the Rothman Index system. Drawing on preliminary findings, hospital stakeholders will be engaged in participatory design to improve workplace protocols and practices around algorithmic decision making. Subsequently, this project will inform three fundamental research challenges: (1) technical communication research on how cognition and memory facilitate workplace writing processes and the impacts of electronic systems on patient-provider and provider-provider communication; (2) organizational research on the synergistic mechanisms that successfully enable VICU; and (3) computer science research on the social and ethical implications of algorithmic interventions such as the Rothman Index. Furthermore, this research will open door to studying many related problems: (a) feasibility of algorithmic patient care for preventative population outreach programs; (b) predictions of job displacements caused by algorithmic patient care implementation and anticipatory workforce training; (c) future algorithmic design for healthcare workplaces.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.