RAPID: Value of Information and Structured Decision-Making for Management of Ebola

在传染病爆发期间，如2014年西非爆发埃博拉病毒，人们创建了许多数学模型来预测疾病的规模、广度和空间范围，并向公共卫生官员通报如何最好地利用和分配资源。根据模型和数据参数的不同，他们可以预测出截然不同的结果。这对政策制定者解释和应用这些工具构成了重大挑战。这个研究小组将利用现有预测模型的优势，并将量化每一种不同模型决策中的不确定性。调查人员将使用目前公布的2014年埃博拉疫情预测模型，模拟其他公共卫生干预措施。这一方法不仅将为当前的疫情提供直接工具，而且还将建立一个比较预测的正式框架。这项研究解决了众多已发表的埃博拉病毒模型的科学不确定性，并将评估每个模型实现其所述目标的能力。研究人员将系统地确定模型中的不确定性领域，以针对管理建议中存在分歧的关键来源。这将利用信息论的价值和完全信息分析的应用价值来实现。该研究团队计划扩大他们的工作范围，在新的埃博拉模型发布后将其包括在内。结果不会选出或产生单一的？最佳？但是，它将阐明并减少阻碍每一种模式取得成功的不确定性。

During an infectious disease outbreak, such as the 2014 outbreak of Ebola virus in West Africa, numerous mathematical models are created to predict the magnitude, breadth, and spatial extent of disease and to inform public health officials on how to best utilize and distribute resources. Depending on the models and the parameters of the data, they can predict vastly different outcomes. This poses a significant challenge for policy-makers to interpret and apply these tools. This research team will utilize the strengths of existing prediction models and will quantify the uncertainty in decision-making for each of the various models. The investigators will simulate alternative public health interventions using current published prediction models for the 2014 Ebola outbreak. This method will not only provide an immediate tool for the current outbreak, but it establishes a formal framework for comparing projections. This research addresses scientific uncertainty across numerous published Ebola virus models and will evaluate the ability of each model to achieve its stated objective. The researchers will systematically identify the areas of uncertainty in the models to target key sources of disagreement in management recommendations. This will be done using value of information theory and applying value of perfect information analysis. The research team plans to expand the scope of their work to include new Ebola models as they become published. The results will not select or produce a single ?best? model, but will rather articulate and reduce the uncertainty that impedes success in each of the models.