基于数据分析的多维不确定环境下手术室调度优化方法研究

In a time when hospitals in China are facing the challenges of operating room (OR) shortage, low utilization levels, and demand surge, effectively improving resource utilization through OR scheduling presents a viable and important approach. However, model development and optimization for OR scheduling are particularly challenging due to the inherent nature of the OR setting - data-driven, multi-resources, multi-dimensional uncertainties, and multi-stakeholders. The current project is undertaken to optimize OR resource utilization, as well as to improve patient and surgical team satisfaction in aspects of data, methods, and application. This project has following key components. First, we identify and estimate key parameters affecting the OR scheduling performance based on medical data analytics. Second, we construct stochastic programming models considering the post anesthesia care unit capacity constraints and surgery process and demand uncertainties. These models use key parameters to address issues such as emergency surgery waiting time target, resource allocation for pre-operation preparation and post-operation cleaning, and cancellation of elective surgeries. Third, through model analysis, we design a two-phase optimization algorithm, an improved multi-objective evolutionary algorithm and a branch-and-price-and-cut algorithm. Lastly, we will focus on real-world research implementations. It is anticipated that this project will make innovative contributions in terms of theory and methods for OR scheduling optimization based on analytics under multi-dimensional uncertainties. We expect this research will help Chinese hospitals improve operating room resource utilization and alleviate the supply-demand imbalance of surgery service.

目前我国医院手术室资源紧缺、利用效率低，与手术服务需求激增之间的供需矛盾突出，通过科学调度提高手术室的资源利用率是缓解这一矛盾的有效途径。手术室调度问题兼具数据驱动性、多资源性、多维不确定性、多参与主体性等难点，对建模与优化方法提出了新挑战。本项目拟从数据、方法和应用三个层面展开，优化资源利用率的同时提高病人、医疗团队的满意度。主要内容包括：基于医疗数据分析，对影响调度性能的关键参数进行科学估计；在术后恢复室容量限制下，针对手术过程和需求不确定性，利用关键参数构建考虑紧急病人手术等待时间阈值、术前准备/术后清理资源配置、择期病人手术取消的随机规划调度模型；通过分析模型性质，分别设计两阶段优化算法、改进多目标进化算法和分支定价切割算法；最后开展应用研究。以期在基于数据分析的多维不确定环境下的手术室调度理论和方法方面取得创新性成果，促进我国医院提高手术室资源利用水平，缓解手术服务的供需矛盾。

手术室是医院的成本和收益中心，手术室调度是医院管理的核心，本项目针对手术室调度的数据驱动性、多资源性、多维不确定性、多参与主体性等难点开展了研究，旨在探索手术室调度的最优理论与方法，以期进一步提升手术室运作效率，保障人民群众的生命健康。取得了如下的研究成果：针对单考虑手术室资源的手术室调度问题，分别构建了紧急病人需求随机、紧急病人手术具有等待阈值和医生手术时间具有偏好的手术调度模型，开发了精确算法和启发式算法；针对考虑术前和术后资源的手术调度问题，分别构建了考虑术前麻醉室和术后病床资源的手术调度模型，设计了精确算法和启发式算法；针对手术室重调度问题，分别构建了考虑行为主体不满意度和非手术室麻醉下的手术重调度模型，在分析模型结构特性基础上，设计了混合智能算法。综合而言，整个研究进行了一些创新型的探索，提出了一些创新型的研究思路，并针对医院手术室管理实践给出了一些手术调度方法和建议。