Using Road Traffic Data to Identify COVID-19 Priority Testing Locations in Southern California

点击翻译按钮获取中文摘要

Project Summary Without a vaccine or effective treatment, there is an urgent need for performing widespread COVID-19 testing to control disease spread. However, complete population testing is prohibitively challenging as testing supplies are limited and require trained health staff which could be better used in caring for those confirmed to be infected. It is therefore critical to focus testing in high-priority areas, where tests are likely to capture positive cases. Identifying infected individuals quickly as tests become more widely available will provide crucial information on overall disease prevalence to inform future disease control efforts. We can help identify areas of potentially high disease prevalence by synthesizing and using traffic patterns, as transportation patterns may shed light on possible transmission patterns in Los Angeles County (LAC). We propose using the USC Archived Data Management System (ADMS), which collects and synthesizes traffic data, to create an epidemic model informed by up-to-date origin-destination traffic information. We will use the model to identify which of the 26 health districts in LAC are at highest risk for unidentified cases and optimally locate testing sites within these regions. This allows our recommendations to incorporate change in transportation patterns as social distancing recommendations evolve. Specifically, we will partner with the LA County Department of Public Health to: 1. Use road sensor data to analyze traffic patterns in Los Angeles County to understand the impact of social distancing guidelines on population flow. 2. Develop a dynamic transmission network model of COVID-19 using results from Aim 1 and disease parameters from the medical literature to identify high priority districts for testing. 3. Develop a location model to optimally place drive-through testing sites in these districts. The proposed work will use methodology from infectious disease transmission models, traffic data, and facility location models together in a novel way. Not only will we provide much needed insight using empirical data into population flow dynamics in the context of social distancing recommendations, we will shed light on infectious disease modeling more generally. By creating a compartmental network model with realistic, time-varying travel patterns in a large metropolitan area, the proposed work will further our understanding of the impacts of structural modeling assumptions on disease prediction.