Maximising the Use of Existing Administrative Data Sets for Wales through Computer Modelling and Simulation

要开展的第一项任务将是与其他利益攸关方协商，并讨论哪些行政数据集需要被视为示范项目。预计将在研究金期限内制定至少两个项目，但如果时间允许，可能还会有其他项目。完成的项目数量将取决于数据的可得性和质量以及所需的模型量。示范项目的选择还将取决于威尔士政府确定哪些政策要求。例如，已经确定了强烈政策要求的两个项目是(1)心脏手术时间表研究，(2)模拟未来的社会护理系统。我预计，该项目的这一领域将由WAG和HMC2的关键人员推动，他们将清楚地知道他们想要开发哪些项目。下一步是查看数据是否可用，或者是否需要获取更多数据。需要就向HMC2提供更多数据集进行谈判。与行政数据有关的任何信息都将被记录在案，以备纳入相关的“关键调查结果”报告。下一步将是审查可能使用的计算机模型和模拟技术。计算机模型和模拟允许在计算机环境中对整个系统进行建模，并运行有助于决策的“假设”情景。例如，可以开发一个计算机模型，表示在特定医院等待心脏手术的患者数量。该模型可以用来观察如果病房增加床位或剧院可用情况发生变化会发生什么。可能使用的建模技术是系统动力学(SD)和离散事件仿真(DES)，这两种技术都已在医疗保健建模(艾滋病毒、结核病和优化医院资源)中大量使用。系统动力学是一种对非常复杂的系统进行建模的方法。它本质上是确定性的，依赖于影响整个系统的反馈回路和时延。可持续发展战略可以提供丰富的质量和数量产出，特别是在政策层面。在离散事件仿真中，系统被建模为一系列事件，其中每个事件在特定时间发生，并导致系统发生变化。DES本质上是随机的，允许不同的事件根据偶然性在系统内发生，从而捕捉到建模系统的不确定性。开发的模型通常非常直观，这有助于将研究结果传达给非数学受众。方法的选择将受到被调查系统的数据和复杂性的影响。通过我在兽医实验室机构的工作和我的博士学位，我有近八年的计算机和模拟模型开发经验，我将利用这些经验来开发示范项目的模型(England等人(2004)，Jones等人(2004)和英格兰人(1997))。一旦计算机模型和模拟技术经过审查并与数学学院和HMC2的专家讨论，示范项目的模型将被开发、测试和验证，以确保它们是所考虑的系统的准确表示。选定的模型将被记录下来，并向其他利益相关者解释。最后，将完成每个项目的“关键发现”和技术报告，并将其提交给该项目的主要利益攸关方。将讨论结论和今后的工作，以及如何将现有数据用于其他目的，以及如何将方法应用于其他领域或提供给其他研究金。我们还希望，该研究金的任何发现都能被撰写出来，并提交给运筹学或其他合适的期刊。

The first task to undertake will be to consult with the other stakeholders and to discuss which administrative data sets need to be considered as demonstration projects. It is envisaged that at least two projects will be developed within the length of the fellowship but there may be others if time allows. The number of projects completed will depend on the availability and quality of the data and the amount of modelling required. The choice of demonstration project will also depend on which policy requirements are identified by the Welsh Government. For example, two projects which have already had strong policy requirements identified are (1) a study of cardiac surgery scheduling, and (2) modelling future social care systems. I envisage that this area of the project will be driven by key personnel in WAG and HMC2 who will have a clear idea about which projects they want to develop. The next stage is to see whether the data is usable or if additional data needs to be acquired. The provision of further data sets to HMC2 would need to be negotiated. Any information relating to the administrative data would be documented ready for its inclusion in the relevant "key findings" report. The next step will be to review possible computer model and simulation techniques that can be used. Computer models and simulation allow whole systems to be modelled in a computer environment, and "what-if" scenarios to be run which help as an aid to decision making. For example, a computer model can be developed which represents the number of patients awaiting cardiac surgery in a particular hospital. The model can be used to see what happens if more beds are added to the ward or if the theatre availability changes. The modelling techniques that are likely to be used are System Dynamics (SD) and Discrete Event Simulation (DES), both of which have been used significantly in healthcare modelling (HIV, TB, and optimising hospital resources). System Dynamics is a way of modelling very complex systems. It is deterministic in nature and relies on feedback loops and time delays which affect the whole system. SD can deliver a wealth of qualitative and quantitative output, particularly at policy level. In Discrete Event Simulation the system is modelled as a sequence of events where each event happens at a specific time and causes a change in the system. DES is stochastic in nature and allows different events to occur within the system according to chance, therefore capturing the uncertainty of the modelled system. The models developed are typically very visual which helps communicate the findings to a non-mathematical audience. The choice of method will be driven by the data and complexity of the system under investigation. Through my work at the Veterinary Laboratories Agency and my PhD, I have almost eight years' experience of developing computer and simulation models which I will draw upon to develop the models for the demonstration projects (England et al (2004), Jones et al (2004) and England (1997)).Once the computer model and simulation techniques have been reviewed and discussed with experts within the School of Mathematics and HMC2, the models for the demonstration projects will be developed, tested and validated to ensure that they are an accurate representation of the systems being considered. The chosen models will be documented and explained to the other stakeholders. Finally, the "key findings" and technical reports for each project will be completed and presented to the main stakeholders in this project. Conclusions and future work will be discussed along with how the available data can be used for other purposes, and how the methodology can be applied to other areas or feed into the other fellowships. It is also hoped that any findings from the fellowship can be written up and submitted to Operational Research or other suitable journals.