Modeling the Geospatial Evolution of COVID-19 using Spatio-temporal Convolutional Sequence-to-sequence Neural Networks
See where this sits in the topic map →Summary AI-generated
- TL;DR
- During the first twelve months of the COVID-19 pandemic, Portugal was severely affected, experiencing three distinct waves and briefly recording the highest incidence rate in the world.
- Problem
- Accurately predicting the geospatial evolution of COVID-19 is challenging because existing analytical methods struggle to capture the complex dynamics of local contagion combined with infections spreading from neighboring regions.
- Method
- Using official municipality-level data from the Portuguese Directorate-General for Health (DGS) for the first year of the pandemic, we generated a sequence of incidence rate maps for mainland Portugal to evaluate different predictive approaches.
- Results
- The modified convolutional sequence-to-sequence neural network outperformed the ARMA, VAR, SIRD, and baseline ConvLSTM models in predicting the spatial-temporal evolution of the incidence rate.
- Contributions
- Not specified in the abstract.
- Limitations
- Not specified in the abstract.
- Takeaways
- Modified convolutional sequence-to-sequence neural networks are more effective at modeling the complex spatial and temporal dynamics of disease spread than traditional statistical, compartmental, and baseline deep learning models.
- Applications
- Not specified in the abstract.
- Topics
- Geospatial Evolution; Spatio-temporal Convolutional Sequence-to-sequence Neural Networks; COVID-19
- For industry
- Healthcare and public health
- Why it matters
- Advances spatio-temporal modeling techniques for tracking and predicting the spread of infectious diseases across geographic regions.
Abstract
Europe was hit hard by the COVID-19 pandemic and Portugal was severely affected, having suffered three waves in the first twelve months. Approximately between January 19th and February 5th 2021 Portugal was the country in the world with the largest incidence rate, with 14-day incidence rates per 100,000 inhabitants in excess of 1,000. Despite its importance, accurate prediction of the geospatial evolution of COVID-19 remains a challenge, since existing analytical methods fail to capture the complex dynamics that result from the contagion within a region and the spreading of the infection from infected neighboring regions. We use a previously developed methodology and official municipality level data from the Portuguese Directorate-General for Health (DGS), relative to the first twelve months of the pandemic, to compute an estimate of the incidence rate in each location of mainland Portugal. The resulting sequence of incidence rate maps was then used as a gold standard to test the effectiveness of different approaches in the prediction of the spatial-temporal evolution of the incidence rate. Four different methods were tested: a simple cell level autoregressive moving average (ARMA) model, a cell level vector autoregressive (VAR) model, a municipality-by-municipality compartmental SIRD model followed by direct block sequential simulation, and a new convolutional sequence-to-sequence neural network model based on the STConvS2S architecture. We conclude that the modified convolutional sequence-to-sequence neural network is the best performing method in this task, when compared with the ARMA, VAR, and SIRD models, as well as with the baseline ConvLSTM model.