Hurricanes result in large scale evacuations almost every year. Of particular concern and difficulty is the decision of whether or not to evacuate hospitals in these emergencies. During an emergency, a hospital is a source of refuge, and evacuating its patients is often viewed as a last resort since it is difficult to provide quality care while transporting them. At the same time, flooding and loss of power and communications put patients and caregivers at very high risk. Most emergency response plans do not have clear guidelines for evacuating or sheltering-in-place. Hurricanes are particularly complicated because there is often considerable uncertainty surrounding their eventual trajectory and intensity. These factors have contributed to, what is in hindsight, poor decisions that have cost lives. The current paper addresses this problem by developing a stochastic optimization formulation, taking into account evolving conditions and, therefore a hopefully robust collection of future flood, wind, and roadway traffic predictions. The model determines the order in which patients should be evacuated over time based on the evolution of the storm by trading off cost and risk. A holistic case study focused on North Carolina and the evolution of Hurricane Isabel is presented by fusing data and model outputs from different sources. The results highlight the advantages of using a recourse formulation that adapts to new information and illustrates the proposed decision-support model’s long-term applications.

飓风几乎每年导致大规模撤离。特别令人关注和困难的是在这些紧急情况下是否撤离医院的决定。在紧急情况下，医院是避难所，而疏散患者通常被视为不得已而为之，因为在运送患者时很难提供优质的医疗服务。同时，洪水，断电和通讯使患者和护理人员处于非常高的风险中。大多数应急响应计划都没有明确的撤离或就地庇护准则。飓风特别复杂，因为飓风的最终轨迹和强度通常存在很大的不确定性。事后看来，这些因素导致了牺牲生命的糟糕决策。当前的论文通过开发一个随机的优化公式来解决这个问题，同时考虑到不断变化的条件，并因此希望对未来的洪水，风和道路交通流量进行预测而收集到有力的信息。该模型根据风暴的演变，通过权衡成本和风险来确定患者随时间撤离的顺序。通过融合来自不同来源的数据和模型输出，提出了一个针对北卡罗来纳州和伊莎贝尔飓风演变的整体案例研究。结果突出显示了使用可适应新信息的资源表述的优势，并说明了所提出的决策支持模型的长期应用。风和道路交通预测。该模型根据风暴的演变，通过权衡成本和风险来确定患者随时间撤离的顺序。通过融合来自不同来源的数据和模型输出，提出了一个针对北卡罗来纳州和伊莎贝尔飓风演变的整体案例研究。结果突出显示了使用可适应新信息的资源表述的优势，并说明了所提出的决策支持模型的长期应用。风和道路交通预测。该模型根据风暴的演变，通过权衡成本和风险来确定患者随时间撤离的顺序。通过融合来自不同来源的数据和模型输出，提出了一个针对北卡罗来纳州和伊莎贝尔飓风演变的整体案例研究。结果突出显示了使用可适应新信息的资源表述的优势，并说明了所提出的决策支持模型的长期应用。