Masterthesis: On mobility for COVID-19 forecasts in Germany using ordinary differential equations and graph neural networks

The network analyis and the commuter-based SIR model was inpired, adapted and extended based on the work of:

"COVID-19 lockdown induces structural changes in mobility networks -- Implication for mitigating disease dynamics", Frank Schlosser, Benjamin F. Maier, David Hinrichs, Adrian Zachariae, Dirk Brockmann, (https://arxiv.org/abs/2007.01583)

All code is self-implemented in R.

The spatio-temporal GNN and the evaluation scheme were adapted and extended from:

"Transfer Graph Neural Networks for Pandemic Forecasting" George Panagopoulos, Giannis, Nikolentzos, Michalis Vazirgiannis, (https://arxiv.org/abs/2009.08388)

The code of this work was re-used and adapted and the original code is available at: https://github.com/geopanag/pandemic_tgnn. This code is located in gnn_experiments/mt-gnn. Specifically in the /code subdirectory.

License

• MIT License