Structural Estimation of Load Balancing Behavior in Inpatient Ward Network
Structural Estimation of Load Balancing Behavior in Inpatient Ward Network
复制标题
住院病房网络负载均衡行为的结构估计
DOI：
--
复制
发表时间：
2020
期刊：
影响因子：
--
通讯作者：
Jing Dong
中科院分区：
文献类型：
--
作者：
Jing Dong
在管理住院病床时，决策者经常面临床位利用率高和需求变化大的问题。通过服务外部署实现负载平衡是缓解拥塞的主要操作手段之一，但这种做法会带来相当大的成本。因此，必须了解决策者在实践中执行负载平衡的程度及其对系统性能的影响。本文采用结构估计方法，实证量化了住院病房占用率对患者路径决策的因果效应。我们利用控制函数方法克服了估计中占用率的内生性问题。我们的结果显示了显著的负载平衡效果，并且在白天和夜间的负载平衡行为之间存在很大差异。基于估计的路由策略，我们进一步研究了负载平衡行为的变化如何影响系统性能。我们的研究结果为管理住院病房网络的拥堵提供了新的视角。我们发现，一天中更一致的负载平衡行为会带来更好的整体系统性能。此外，我们发现允许基于底层网络结构的不同非服务病房的异构负载平衡因素为医院管理者提供了一种低成本和低风险的方式来减少网络中最繁忙病房的拥塞。
In managing inpatient ward beds, decision makers often face high levels of bed utilization and large variability in demand. Load balancing via off-service placement is one of the primary operational levers to alleviate congestion, yet this practice is associated with considerable costs. Therefore, it is imperative to understand how much decision makers exercise load balancing in practice and its implications on system performance. In this paper, we take the structural estimation approach and empirically quantify the causal effect of inpatient wards’ occupancy on patient routing decisions. We leverage the control function method to overcome the endogeneity problem of occupancy in estimation. Our results reveal a significant load balancing effect, and a large difference between the load balancing behavior during daytime and nighttime. Based on the estimated routing policy, we further study how changes to the load balancing behavior affect system performance. Our findings provide new perspectives to manage congestion in the inpatient ward network. We find that more consistent load balancing behavior throughout the day leads to overall better system performance. Moreover, we find that allowing for heterogeneous load balancing factors for different off-service wards based on the underlying network structure provides hospital managers with a low cost and low risk way to reduce congestion in the busiest wards in the network.