考虑患者特征及下游资源限制的鲁棒手术调度  

Robust surgery scheduling with patient feature and downstream resource constraints

作　　者：王昱 张舒政 曲刚 WANG Yu;ZHANG Shuzheng;QU Gang(School of Business Administration,Northeastern University,Shenyang 110819,China;Health Commission of Dalian,Dalian 116001,China)

机构地区：[1]东北大学工商管理学院,沈阳110819 [2]大连市卫生健康委员会,大连116001

出　　处：《系统工程理论与实践》2025年第1期192-207,共16页Systems Engineering-Theory & Practice

基　　金：国家自然科学基金(72071036,72272029)。

摘　　要：患者的身体状况具有异质性和复杂性,在术前往往难以准确预估患者的手术及术后恢复时长,为医院制定手术调度方案带来了挑战.医疗信息化产生了大量医疗数据,有效利用历史相似患者的诊疗数据,可以为新到达患者的手术调度提供指导.本文考虑了患者手术时长和术后恢复时长的不确定性,基于机器学习方法构建了患者特征驱动的不确定集.以最小化调度周期内运营成本为目标,建立患者细分视角下的手术资源协同调度两阶段鲁棒优化模型,并设计列与约束生成算法进行精确求解.利用医院现实数据进行数值实验,结果表明患者特征驱动的鲁棒优化方法可以提高手术调度方案的质量,并缓解下游床位的短缺现象.The physical condition of patients is heterogeneous and complex,making it challenging to accurately predict surgery duration and postoperative recovery length before surgery,which brings challenges to the hospital in surgery scheduling.Medical informatization has produced a large amount of medical data,and the effective use of the diagnosis and treatment data of history patients can provide guidance for the surgery scheduling of newly arrived patients.This paper proposes a machine learning approach that constructs a patient feature-based uncertainty set,taking into account the uncertainty of surgery duration and postoperative recovery length.To minimize total costs during the scheduling period,a two-stage robust optimization model is established from the perspective of patient segmentation,and a column and constraint algorithm is designed to generate accurate solutions.Numerical experiments with real hospital data show that the proposed method can improve the quality of surgical scheduling and alleviate downstream unit shortages.

关 键 词：手术调度 鲁棒优化 机器学习 医疗运作管理 

分 类 号：O224[理学—运筹学与控制论] F224[理学—数学]