BlueDot, 207 Queens Quay West #801b, Toronto, Ontario, Canada.
BlueDot, 207 Queens Quay West #801b, Toronto, Ontario, Canada.
Spat Spatiotemporal Epidemiol. 2021 Feb;36:100380. doi: 10.1016/j.sste.2020.100380. Epub 2020 Oct 27.
Air travel is an increasingly important conduit for the worldwide spread of infectious diseases. However, methods to identify which airports an individual may use to initiate travel, or where an individual may travel to upon arrival at an airport is not well studied. This knowledge gap can be addressed by estimating airport catchment areas: the geographic extent from which the airport derives most of its patronage. While airport catchment areas can provide a simple decision-support tool to help delineate the spatial extent of infectious disease spread at a local scale, observed data for airport catchment areas are rarely made publicly available. Therefore, we evaluated a probabilistic choice behavior model, the Huff model, as a potential methodology to estimate airport catchment areas in the United States in data-limited scenarios. We explored the impact of varying input parameters to the Huff model on estimated airport catchment areas: distance decay exponent, distance cut-off, and measures of airport attractiveness. We compared Huff model catchment area patterns for Miami International Airport (MIA) and Harrisburg International Airport (MDT). We specifically compared our model output to observed data sampled for MDT to align model parameters with an established, observed catchment area. Airport catchment areas derived using the Huff model were highly sensitive to changes in model parameters. We observed that a distance decay exponent of 2 and a distance cut-off of 500 km represented the most realistic spatial extent and heterogeneity of the MIA catchment area. When these parameters were applied to MDT, the Huff model produced similar spatial patterns to the observed MDT catchment area. Finally, our evaluation of airport attractiveness showed that travel volume to the specific international destinations of interest for infectious disease importation risks (i.e., Brazil) had little impact on the predicted choice of airport when compared to all international travel. Our work is a proof of concept for use of the Huff model to estimate airport catchment areas as a generalizable decision-support tool in data-limited scenarios. While our work represents an initial examination of the Huff model as a method to approximate airport catchment areas, an essential next step is to conduct a quantitative calibration and validation of the model based on multiple airports, possibly leveraging local human mobility data such as call detail records or online social network data collected from mobile devices. Ultimately, we demonstrate how the Huff model could be potentially helpful to improve the precision of early warning systems that anticipate infectious disease spread, or to incorporate when local public health decision makers need to identify where to mobilize screening infrastructure or containment strategies at a local level.
航空旅行是传染病在全球范围内传播的一个越来越重要的途径。然而，识别个人可能使用哪些机场开始旅行，或者个人到达机场后可能前往哪些地方的方法还没有得到很好的研究。通过估计机场服务区可以解决这一知识空白：机场的客源地理范围。虽然机场服务区可以提供一个简单的决策支持工具，帮助划定传染病在当地范围内传播的空间范围，但机场服务区的观测数据很少公开提供。因此，我们评估了一个概率选择行为模型，即 Huff 模型，作为一种在数据有限的情况下估计美国机场服务区的潜在方法。我们探讨了改变 Huff 模型输入参数对估计机场服务区的影响：距离衰减指数、距离截止值和机场吸引力的度量。我们比较了迈阿密国际机场（MIA）和哈里斯堡国际机场（MDT）的 Huff 模型服务区模式。我们特别将模型输出与为 MDT 采样的观测数据进行比较，以使模型参数与已建立的观测服务区保持一致。使用 Huff 模型得出的机场服务区对模型参数的变化非常敏感。我们观察到，距离衰减指数为 2，距离截止值为 500 公里，代表了 MIA 服务区最现实的空间范围和异质性。当将这些参数应用于 MDT 时，Huff 模型产生的空间模式与观测到的 MDT 服务区相似。最后，我们对机场吸引力的评估表明，与所有国际旅行相比，前往传染病输入风险特定国际目的地（即巴西）的旅行量对预测机场选择的影响很小。我们的工作是使用 Huff 模型作为数据有限情况下的通用决策支持工具来估计机场服务区的概念验证。虽然我们的工作代表了对 Huff 模型作为一种近似机场服务区的方法的初步考察，但下一步是根据多个机场对模型进行定量校准和验证，可能需要利用本地人类流动数据，例如来自移动设备的通话记录或在线社交网络数据。最终，我们展示了 Huff 模型如何有助于提高预测传染病传播的预警系统的精度，或者在地方公共卫生决策者需要确定在地方一级在哪里调动筛查基础设施或控制策略时，如何纳入该模型。
