RAPID: development of a local epidemiological population balance model informed by UAV and WVD data

大学和周边地区的决策和政策制定需要了解人们如何在环境中移动和相互作用。该RAPID项目采用人口平衡建模方面的新科学方法，对大学校园和周边城镇的人员流动和互动进行建模，目的是提供新工具，帮助制定减轻和最终消除新型冠状病毒以及未来生物威胁的合理战略。模型输入的数据将通过人工智能算法分析公共、户外区域（包括绿地/公园、人行道/街道和校园人行道/聚集空间）的高清视频片段获得。最初为研究复杂流体而开发的高效工具将能够确定流行病学模型所需的关键参数，包括有效传播率。流行病学模型将被转化为一个仪表板，供决策者使用，并用于有关缓解战略的公众教育。该RAPID项目将提供一种计算工具和范例，供社区更广泛地使用，并在其他和未来具有挑战性的公共卫生问题上使用。应用于高校和地方自治市的多元人口平衡模型将生成基于主体的流行病学模型的关键参数。多元平衡建模将面临新的局部人口密度和运动数据集的挑战，该模型参数估计使用在先前和当前NSF支持下开发的并行回火。除了通常的免疫、易感、暴露、感染和恢复类别的区分之外，需要考虑的其他变量包括：年龄，特别是与大学生相关的年龄，面部覆盖，室内和室外，以及实时更新空中（无人机）和地面（可穿戴视频设备增强的固定摄像头）监控数据提供的时空人口分布。虽然在大规模流行病学模型中包含交通网络提供的粗粒度信息是很常见的，但本项目将探索社会力量模型与流行病人口平衡模型相结合所提供的机会。先进的并行调节算法将在GPU集群上运行，以每日数据流挑战模型，以更新流行病学模型和情景预测的参数。将为政策制定和公众教育提供一个项目仪表板。更广泛的影响包括可应用于广泛公共卫生问题的计算工具。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Decision making and policy setting by universities and surrounding localities requires knowledge of how people move and interact in the environment. This RAPID project adapts new scientific approaches in population balance modeling to model human movement and interaction on a university campus and a surrounding town with the goal of providing new tools to help develop rational strategies for mitigation and eventual elimination of the novel corona virus, as well as future biological threats. Data for the model input will be obtained from high-definition video footage of public, outdoor areas including green spaces/parks, sidewalks/streets, and campus walkways/congregating spaces analyzed by artificial intelligence algorithms. Highly efficient tools that were originally developed to study complex fluids will enable determination key parameters needed for epidemiological models including effective transmission rates. Epidemiological modeling will be translated into a dashboard for use by policy makers as well as for public education about mitigation strategies. This RAPID project will provide a computational tool and example for use more broadly by communities and in additional and future, challenging public health issues.A multivariate population balance model applied to a college and local municipality will generate key parameters for agent-based epidemiological models. Multivariate balance modeling will be challenged with new data sets of local population density and motion for model parameter estimation using parallel tempering developed under prior and current NSF support. In addition to the usual distinctions of immune, susceptible, exposed, infected, and recovered classes, additional variables to consider include: age, especially relevant for University students, face-covering, inside and outside, and spatial-temporal population distributions afforded by real time updates of aerial (unmanned aerial vehicle) and ground (stationary camera augmented by wearable video devices) surveillance data. While it is common to include coarse-grained information afforded by transportation networks in large-scale epidemiology models, this project will explore opportunities afforded by social force models combined with epidemic population balance modeling. Advanced parallel tempering algorithms will be run on a GPU cluster to challenge the model with daily data streams to update parameters for epidemiological models and scenario projections. A project dashboard will be made available for policy decision making and public education. Broader impacts include computational tools that can be applied to a broad range of public health issues.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.