Reinforcement Learning Methodology for Decision Analysis and Support in Long-term Care

我们拟议的工作将与我们的合作伙伴加拿大软件公司PointClickCare一起开发基于强化学习的新数据分析方法，该方法将用于构建新的分析工具，以更好地支持长期护理行业的决策。长期护理中的居民的健康结果受到床边个人决策的影响，以及他们接受护理的设施管理员做出的设施级别决策的影响。该项目的目标是开发能够使用健康记录和行政数据的数据分析方法，以更好地了解个人和设施一级的哪些决策战略与长期护理居民和工作人员的最佳结果有关，重点是传染病。这些新的数据分析方法将应用于来自长期护理设施的数据以及关于传染病暴发的数据，以便开发软件工具，通过考虑关于居民、设施和这些设施运行的背景的信息，帮助用户了解个人和设施层面的哪些决策策略是最有效的。由此产生的工具将被整合到PointClickCare的电子健康记录产品中，并向他们在加拿大和国际上的客户提供，使他们能够更好地为他们的决策提供信息，并为他们的居民和工作人员改善结果。

Our proposed work will develop new data analysis methodology based on reinforcement learning with our partner, the Canadian software company PointClickCare, which will be used to construct new analysis tools to better support decision-making in the long-term care industry.The health outcomes of residents in long-term care are affected by individual decisions made at the bedside, as well as by facility-level decisions made by administrators at the facilities where they receive care. The goal of this project is to develop data analysis methods that can use health record and administrative data to better understand what decision-making strategies, both at the individual and the facility level, are associated with the best outcomes for long-term care residents and staff, with a focus on infectious diseases. These new data analysis methods will be applied to data from long-term care facilities together with data about infectious disease outbreaks in order to develop software tools that help users understand what decision strategies at the individual and the facility levels are most effective, by taking into account information about residents, facilities, and the context where those facilities operate. The resulting tools will be incorporated into PointClickCare's electronic health record products and made available to their clients in Canada and internationally to allow them to better inform their decision-making and improve outcomes for their residents and staff.