Analyzing the relationship between COVID-19 infections and human mobility
September 28, 2021
I am Hoeyun, a fourth-year Ph.D. student in the Geo-social lab. I was so happy to present our research at GIScience AMD’21 which was held virtually on September 27, 2021 on behalf of my co-authors, Kaitlyn Hom, Mark Rifkin, Beichen Tian, and Caglar Koylu. The purpose of our research is to examine the spatiotemporal heterogeneity in the relationship between human mobility and COVID-19 cases. We compare two time series for each county in the U.S., one for the daily number of COVID-19 cases and the other one for daily human mobility flows into and within each county. The method we use is dynamic time warping (DTW) that calculates a similarity using a distance between two time series. The image below shows the map of DTW distances and the time series of six distinct areas, which shows clear variations across space and time. Even in the same metropolitan areas, Los Angeles, Austin, Texas, and New York City have low degrees of similarity, while Chicago, St. Louis, and Washington DC have high degrees of similarity. In New York City, for instance, COVID-19 cases keep increasing regardless of the trend of mobility flow. In Washington DC, however, we can see the patterns of two time series are pretty similar to each other. Also, even in the same counties, there are temporal heterogeneities. In Los Angeles, for example, during the early period, it seems like the COVID-19 trend is following the mobility flow trend, but after that, the COVID-19 trend is leading the trend of mobility flow. So, we could conclude that the relationships between human mobility and COVID-19 infections are heterogeneous both spatially and temporally. This study shows the preliminary results of our ongoing project, and there will be more coming soon!
Click here for the AMD'21 workshop information.
Check the figure from our paper:
Kwon, H., Hom, K., Rifkin, M., Tian B. & Koylu, C. (2021). Exploring the spatiotemporal heterogeneity in the relationship between human mobility and COVID-19 prevalence using dynamic time warping. GIScience 2021 Workshop on Advancing Movement Data Science (AMD’ 2021), September 27, 2021, World Wide Web. DOI: https://arxiv.org/abs/2109.13765