The outbreak of SARS-CoV-2 and the corresponding surge in patients with severe symptoms of COVID-19 put a strain on health systems, requiring specialized material and human resources, often exceeding the locally available ones. Motivated by a real emergency response system employed in Northern Italy, we propose a mathematical programming approach for rebalancing the health resources among a network of hospitals in a large geographical area. It is meant for tactical planning in facing foreseen peaks of patients requiring specialized treatment. Our model has a clean combinatorial structure. At the same time, it considers the handling of patients by a dedicated home healthcare service, and the efficient exploitation of resource sharing. We introduce mathematical programming heuristic based on decomposition methods and column generation to drive very large-scale neighborhood search. We evaluate its embedding in a multi-objective optimization framework. We experiment on real world data of the COVID-19 in Northern Italy during 2020, whose aggregation and post processing is made openly available to the community. Our approach proves to be effective in tackling realistic instances, thus making it a reliable basis for actual decision support tools.

SARS-CoV-2 的爆发以及相应的 COVID-19 严重症状患者激增给卫生系统带来了压力，需要专门的物质和人力资源，而这些资源往往超过了当地可用的资源。受意大利北部采用的真实应急响应系统的启发，我们提出了一种数学规划方法，用于在较大地理区域的医院网络之间重新平衡卫生资源。它用于战术规划，以应对可预见的需要专门治疗的患者高峰。我们的模型有一个干净的组合结构。同时，它考虑了通过专门的家庭医疗保健服务来处理患者，以及资源共享的有效利用。我们引入了基于分解方法和列生成的数学规划启发式来驱动超大规模的邻域搜索。我们评估它在多目标优化框架中的嵌入。我们在 2020 年对意大利北部 COVID-19 的真实世界数据进行了实验，这些数据的聚合和后处理向社区公开。我们的方法被证明可以有效地处理现实情况，从而使其成为实际决策支持工具的可靠基础。