- Constantine Kontokosta (Principal Investigator) – Industry Associate Professor of Urban Science and Associated Faculty at CUSP
- Arpit Gupta (Co-Principal Investigator) – NYU Stern School of Business
- Lorna Thorpe (Co-Principal Investigator) – NYU Grossman School of Medicine
A RAPID project recently funded by the NSF will marshal large-scale, anonymized mobility data to estimate variations in exposure density, measure and monitor the effectiveness of social distancing, and predict the spread of the disease based upon mobility patterns.
The research will focus on understanding socioeconomic disparities in outbreak progression and the impacts of shelter-in-place mandates. Using these models, public health officials will be able to estimate the likelihood of successful containment efforts for specific localities and predict where future localized outbreaks and chains of transmission may emerge. The study team also hopes to use the data to identify at-risk communities based on predicted movement patterns and neighborhood socioeconomic characteristics influencing residents’ vulnerability.
Constantine Kontokosta leads the team; he is an associated faculty member in the Civil and Urban Engineering Department and CUSP, and is an associate professor of Urban Science and Planning at the NYU Marron Institute, as well as director of the Urban Intelligence Lab and director of Civic Analytics. Co-principal investigators are Arpit Gupta, an assistant professor of finance at the Leonard N. Stern School of Business, and Lorna E. Thorpe, a professor of population health at the NYU Grossman School of Medicine.
The disasterous spread of COVID-19 has highlighted both the growing global risk of emerging pandemics and the urgent need for enhanced data-driven tools to identify, contain, and mitigate their effects, particularly in dense urban areas. Large-scale, high-resolution data on mobility, travel behavior, and proximate contacts can be used to enhance computational modeling of chains of transmission and future infection outbreaks and localized at-risk “hotspots”. However, current efforts using ad hoc and manual collection of contact and travel data are time and resource consuming, and subject to recollection errors and incomplete histories. Given significant diversity in local, regional, and international movement patterns, simple assumptions of homogenous mobility behaviors, like those typically used in agent-based spatial epidemiological models, can add uncertainty to predictions of transmission risk and outbreak progression. To address these limitations and advance the science and practice of COVID-19 computational modeling, the study team will use a large-scale geolocation dataset, provided by VenPath, Inc., derived from smart phone location information from over 200 applications for approximately 60 million unique users across the United States.
This project will develop computational models derived from the high-resolution mobility data to (1) estimate the contact density for individual users and localities, (2) monitor and measure the extent and effectiveness of “social (physical) distancing” efforts, and (3) predict the geographic extent of disease spread based on movement and travel patterns for individual communities. Using these models, public health officials will be able to estimate the likelihood of successful containment efforts for specific localities and predict where future localized outbreaks and chains of transmission may emerge. Furthermore, the study team proposes to demonstrate how these data can be used in public health communication during disease outbreaks by identifying “at-risk” communities based on predicted movement patterns from specific neighborhoods and neighborhood socioeconomic characteristics influencing residents’ vulnerability. The data processing workflow and mobility models proposed can provide a new resource to epidemiologists and public health officials as they plan, implement, and evaluate the full range of policy responses to infectious disease outbreaks. Specifically, Objective 1 enables officials and researchers to understand the potential transmission rate in a given locality based on the estimated contact density derived from contact tracing using smartphone data. Objective 2 allows for both a real-time and ex post (assuming access to current data) evaluation of social distancing measures and their relative effectiveness in minimizing further spread of infection. Objective 3 creates a methodology to predict the spatio- temporal patterns of disease outbreak and identify “at-risk” locations based on the estimated contact density and mobility trajectories for areas were infections have been identified. This research is intended to provide proof-of-concept locational data analysis tools. It is focused on foundational data processing and model building to support public health decision-making. It is also designed to enable a rapid scale-up in capabilities and operational decision-making using current locational data as actual infection locations are identified. Furthermore, the study team will develop a visualization tool to support web-based analysis of the developed models and metrics. This tool will be made available to public health officials and other decision-makers. Beyond the current threat posed by COVID-19, the tools developed in this project could be applied to a full range of emergency situations, such as hurricanes, flooding, or other natural disasters.