CAREER: Data-driven Models of Human Mobility and Resilience for Decision Making

该项目设想了移动网络物理系统（CPS），人们携带手机产生大量的位置信息，用于感知、计算和监测人类在环境错位期间与物理环境的相互作用。主要目标将是确定人群对特定类型冲击的反应类型，为决策者提供准确和信息丰富的数据驱动的陈述，他们可以利用这些陈述来制定准备和应对计划。此外，该项目的成果将允许开发工具，通过移动CPS的反馈循环，评估和提高不同类型的准备和应对政策的有效性。这些反馈循环可以显示，当基于CPS的计算重新定义政策时，社区在冲击期间的行为是如何变化的，反之亦然。PI和其他人之前的工作已经表明，CPS将人和手机集成为传感平台，可用于大规模收集位置信息，并使用数据挖掘和机器学习技术计算人类在冲击期间的移动行为。然而，大多数结果都是非常有限和临时的，缺乏任何类型的准备和响应策略的严重适用性。该项目将通过开发准确的方法和有效的工具，在移动CPS冲击期间进行决策，从而推动最新技术的发展。从更广泛的影响角度来看，拟议的研究将在两个领域做出贡献：(a)实际部署，促进数据驱动的政策制定，数据驱动的人类行为分析，以及在移动CPS中使用反馈回路进行决策评估；(b)在数据科学领域制定教育计划和培训机会，以促进社会公益，并为决策制定移动CPS。该项目的主要成果将包括用于移动CPS的新型数据驱动方法，这些方法将可靠地描述和预测人类在冲击期间的流动模式和复原力，从而改进准备和响应政策。该项目将利用手机元数据和社交媒体实现以下三个目标：(1)利用移动CPS的实时数据来描述社区对不同类型冲击的反应类型，这将有助于制定更充分的防范政策，为未来的事件做好准备；(2)利用CPS反馈回路中的人类行为信息（实时或批量处理），建立预测方法，预测冲击管理政策对冲击期间人类流动行为和社区恢复力的影响；(3)评估移动CPS中不同冲击、时空尺度和数据源的反应类型和预测方法的可转移性，为决策者分析CPS中手机元数据不完全可用的社区的行为和恢复力提供可能。从知识价值的角度来看，所提出的方法将推动智能和互联社区领域CPS数据分析和实时系统的最新发展。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。

This project envisions mobile cyber-physical systems (CPS) where people carrying cell phones generate large amounts of location information that is used to sense, compute and monitor human interactions with the physical environment during environmental dislocations. The main objective will be to identify the types of reactions populations have to a given type of shock, providing decision makers with accurate and informative data-driven representations they can use to create preparedness and response plans. Additionally, the outcomes of this project will allow for the development of tools to assess and improve the effectiveness of different types of preparedness and response policies through feedback loops in the mobile CPS. These feedback loops could show how community behaviors during shocks change when policies are re-defined based on the computations of the CPS, and vice-versa. Previous work by the PI and others has already showed that CPS integrating people and cell phones as sensing platforms can be used to collect location information at large scale and to compute, using data mining and machine learning techniques, human mobility behaviors during shocks. However, most of the results are very limited and ad-hoc, lacking any type of serious applicability from a preparedness and response policy. This project will advance the state of the art by developing accurate methods and effective tools for decision-making during shocks in mobile CPS. From a broader impacts perspective, the proposed research will contribute in two areas: (a) real-world deployments, to promote data-driven policy development, data-driven analyses of human behavior, and the use of feedback loops in mobile CPS for decision-making assessment; and (b) the creation of an educational plan and training opportunities in the areas of data science for social good and mobile CPS for decision making.The main outcomes of the project will include novel data-driven methods for mobile CPS that will reliably characterize and predict human mobility patterns and resilience during shocks so as to improve preparedness and response policies. The project will make use of cell phone metadata and social media to achieve the following three objectives: (1) to characterize the types of reactions that communities have to different kinds of shocks using real-time data from mobile CPS, which would allow for the development of more adequate preparedness policies to be ready for future events; (2) to create predictive methods to forecast the impact that shock management policies would have on human mobility behaviors and community resilience during a shock, using human behavioral information from the CPS feedback loop when different policies are applied (either in real-time or in batch processing); and (3) to evaluate the transferability of the types of reactions and predictive methods across different shocks, spatio-temporal scales and data sources in mobile CPS, which would provide decision makers with the possibility of analyzing behaviors and resilience in communities where cell phone metadata in the CPS is not fully available. From an intellectual merit perspective, the proposed methods will advance the state of the art in data analytics and real-time systems for CPS in the area of Smart and Connected Communities.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.