A data analytics framework for the application of pedestrian dynamics to public health - Project Summary: The local movement and interaction patterns of individuals congregated in public locations, such as entertainment venues and transportation hubs, impacts public health in myriad ways. For instance, infectious disease transmission in crowded areas, such as the 2016 measles outbreak in Disney world that resulted in 125 cases, is affected by the evolution of the pedestrian contact network. In such contexts, mathematical models can be used to explore different “what if” scenarios for planning public health interventions. For example, pedestrian mobility models could help in the design of public spaces and policies that reduce contacts to mitigate disease spread or encourage walking to improve health outcomes. However, conventional models either ignore human movement or rely on coarse-scale features of human movement, which limits the types of public health interventions that can be attempted. Understanding the fine-scale movement and interaction patterns of people can help design effective policies and spatial layouts to better engineer suitable movement and interaction patterns for improved public health outcomes in several domains. Particle dynamics provides a modeling methodology to study movement and interactions from the cellular to the population level. Our long-term goal is to create a data-driven computational infrastructure for particle dynamics, targeting public health applications. The objective of this proposal is to develop and evaluate an application-agnostic pedestrian dynamics model, which considers each human to be a particle. Here, we will simulate the fine-scaled movement of people in several critical public health contexts. In preliminary work, we have shown that such models can accurately simulate fine-scale movement patterns on a small-scale, yielding effective policies to reduce infection spread while boarding planes and moving in queues. However, directly scaling this to more complex situations, such as an entire theme park, is difficult due to inherent variability in human movement patterns. In this proposal, we will apply novel data sources, such as location-based services (from cell phone apps) to augment such models. Our central hypothesis is that combining location-based service (LBS) data with pedestrian dynamics modeling can uncover movement patterns of people in complex situations with many public health applications. In Aim 1, we will develop an application-agnostic pedestrian dynamics modeling framework that assimilates LBS data. We will compare our approach to methods that do not utilize LBS in order to evaluate accuracy of human movement across multiple scenarios. In Aim 2, we will apply the pedestrian movement and interaction information to a variety of public health domains. These include: viral infection spread at local and global scales, enhancing walkability for active aging, and safe evacuation of the elderly. Finally, in Aim 3, we will translate our pedestrian dynamics modeling framework into public health practice. We will provide our platform to different stakeholders and obtain feedback on user satisfaction to improve the system design.
