STAGCN: Spatial-Temporal Attention Based Graph Convolutional Networks for COVID-19 Forecasting

Nevasini Sasikumar, Krishna Sri Ipsit Mantri

Keywords: covid-19 forecasting, attention, graph neural networks, spatio-temporal, graph convolution, graph modeling

TL;DR: We propose to consider the proximity of infection hotspots into accout when forecasting covid-19 infections. To this end, we model the problem on graphs, and propose spatio-temporal attention based GCN to solve this problem.

Abstract:

The recent outbreak of the novel coronavirus known as the COVID-19 pandemic
has harmed the lives of millions of people across the globe and has imposed a sig-
nificant threat to global healthcare due to its severe transmission capacity. It’s of
utmost importance to be able to accurately forecast the COVID-19 pandemic and
to provide the necessary precautionary measures to protect the health of individu-
als and prevent the spread of this deadly widespread virus. In this paper, we pro-
pose to forecast the upcoming newly infected patients that are likely to be affected
by COVID-19 in prior using a novel deep learning framework, Spatio-Temporal
Attention Based Graph Convolution Networks (STAGCN) to effectively make use
of spatial and temporal relationships. Instead of using traditional time-series fore-
casting techniques at a single city using raw data, we model the problem using
graphs and aim at taking into account the dependency that an infection in one city
has on its neighbors. Our experiments show that STAGCN effectively captures
this dependency and consistently outperforms the other conventional methods.