数据驱动的医疗服务系统资源分配方法研究

本项目利用数字化医疗平台、医疗大数据等信息技术成果，结合基于主体的仿真模型的准确性与离散事件模型的有效性设计仿真优化方法帮助医院实现医疗资源的高效优化配置。本项目将建立综合考虑医院运营成本与患者等待时间的最优资源分配问题；开发实际医疗数据驱动下基于主体的医疗服务系统仿真模型；提出基于双保真模型的仿真优化方案来分配医生、护士、设备等医疗资源到服务流程的各个环节。本项目将开发分布敏感性的广义似然比估计，进而提出基于梯度的仿真最大似然估计将数据驱动的概念推广至复杂随机模型,从而在减小仿真计算代价的同时更高效准确地做出最优决策。

This project leverages both the accuracy of the agent-based simulation model and the efficiency of discrete events systems to develop simulation optimization technique for helping optimize the healthcare resource allocation for hospitals, utilizing the recent advances in the digital healthcare platform and medical big data. We will establish an optimal resource allocation problem considering both the operational cost of the hospital and the waiting time of patients; we will develop a data-driven agent-based simulation model for the healthcare service system; we will propose a simulation optimization technique based on bi-fidelity models to allocate healthcare resource including doctors, nurses, and equipment to different stages of the service process. Our project will develop the generalized likelihood ratio method for distribution sensitivities and then the gradient-based simulated maximum likelihood estimation, which extends the data-driven concept to complex stochastic models, for reducing the computational cost and simultaneously making the optimal decision efficiently.