Spatiotemporal prediction of COVID-19 cases using inter- and intra-county proxies of human interactions

Behzad Vahedi,Morteza Karimzadeh,Hamidreza Zoraghein

doi:10.1038/s41467-021-26742-6

Behzad Vahedi, Morteza Karimzadeh + Show 1 more

Open Access

https://doi.org/10.1038/s41467-021-26742-6

Copy DOI

Abstract

Measurements of human interaction through proxies such as social connectedness or movement patterns have proved useful for predictive modeling of COVID-19, which is a challenging task, especially at high spatial resolutions. In this study, we develop a Spatiotemporal autoregressive model to predict county-level new cases of COVID-19 in the coterminous US using spatiotemporal lags of infection rates, human interactions, human mobility, and socioeconomic composition of counties as predictive features. We capture human interactions through 1) Facebook- and 2) cell phone-derived measures of connectivity and human mobility, and use them in two separate models for predicting county-level new cases of COVID-19. We evaluate the model on 14 forecast dates between 2020/10/25 and 2021/01/24 over one- to four-week prediction horizons. Comparing our predictions with a Baseline model developed by the COVID-19 Forecast Hub indicates an average 6.46% improvement in prediction Mean Absolute Errors (MAE) over the two-week prediction horizon up to 20.22% improvement in the four-week prediction horizon, pointing to the strong predictive power of our model in the longer prediction horizons.

Highlights

Measurements of human interaction through proxies such as social connectedness or movement patterns have proved useful for predictive modeling of COVID-19, which is a challenging task, especially at high spatial resolutions
We demonstrated that incorporating (1) spatiotemporal lags using intercounty indices of connectedness and (2) intracounty measurements of movement improves the performance of highresolution COVID-19 predictive models, especially over longterm horizons
By incorporating the aforementioned spatiotemporal lags, the SpatioTemporal XGB (STXGB)-FB and STXGB-SG models outperformed the COVIDhub-Baseline model in two, three, and four-week prediction horizons on average, with inconsistent comparisons in the one-week horizon

Summary

Introduction

Measurements of human interaction through proxies such as social connectedness or movement patterns have proved useful for predictive modeling of COVID-19, which is a challenging task, especially at high spatial resolutions. We capture human interactions through 1) Facebook- and 2) cell phone-derived measures of connectivity and human mobility, and use them in two separate models for predicting county-level new cases of COVID-19. The underlying assumption in this approach is that more movements between spatial units lead to higher interactions, and an elevated risk of disease spread It is unclear, which of these approaches—using social-media connectedness versus cell-phone-derived humanmobility flow—is a better indicator of physical interaction within and between different regions. The underlying assumption in each approach may not necessarily be valid in the case of COVID-19: considering the sporadic and regional stay-athome orders across the United States, social connectedness may not lead to physical interaction, at least not to the same level as prepandemic. Given the recommended preventive measures such as mask-wearing and physical distancing[13], human flow from one location to another may not necessarily lead to physical interactions that could communicate the disease, especially in public places, where preventive measures are enforced more strictly

Objectives

Methods

Results

Conclusion