在健康中国战略的指引下，分级诊疗制度建设已成为新医改工作的重中之重；而医联体作为推进分级诊疗制度的重要抓手，在实际运作过程中仍存在显著问题。本项目基于延安医疗集团医联体的实际数据，运用数据驱动方法，对医联体的服务运营管理与优化进行系统性研究。首先，运用关联规则等技术，对患者就医机构选择行为进行分析，提出患者的基层首诊促进策略。其次，运用层次聚类分析和贝叶斯分类器等方法，识别患者的就医机构选择行为模式，构建上/下转诊指征和转诊模型；同时构建考虑转诊率的三级医院门诊预约调度优化模型。最后，基于患者基层首诊比例和上/下转诊率，运用数据挖掘技术，对基层医院和三级医院的医生和医疗设备的需求进行预测，构建医生和医疗设备的调度优化模型。项目综合运用数据挖掘、运筹学、分布式鲁棒优化及运营管理理论与方法，研究结果可为医联体的运营管理和资源优化决策提供可靠的理论依据和决策。

Guided by the healthy China strategy, the construction of hierarchical medical system has become the key of the new medical reform. As an important grasper of promoting hierarchical medical system, the medical alliance' operations management also has some problems. Basing on the big-data of Yan'an Medical Group medical alliance, we use data-driven methods to study the service operations management and optimization of the medical alliance systematically. Firstly, though association rule and so on, we analyze the patients' medical institution choice behaviors, and propose the promotion strategy of patient primary diagnosis. Secondly, through the hierarchical clustering and Bayesian classifier, we identify the patients' medical institution choice behavior patterns. Then we construct the index characteristics of dual referral and the referral models. Basing on the referral rate, we construct the optimization model of outpatient appointment scheduling with consideration dual referral. Finally, basing on the ratio of patient primary diagnosis and the rate of dual referral, and using the data-mining technology, we predict the demands of doctors and medical equipment in primary and tertiary hospitals. Then we construct the scheduling optimization models of the doctor and medical equipment respectively. We combine data mining, operations research, distributionally robust optimization, operational management theory and methods. Our research results will provide reliable theoretical basis and decision support for the operations management and optimization of medical alliances.