Video Presentation: https://youtu.be/OhdPB70GQkc
Google Slides Link: https://docs.google.com/presentation/d/1-cclOln9xES9shQd0WSOjhqnX1IcIQzH61doUMB9-jI/edit?usp=sharing
Disease can spread quickly and quietly, and countermeasures are often reactive, waiting for confirmed cases to appear. To predict the spread of disease is an advantage that can protect economies and lives.
Global pandemics merit global models. However, the detail, recency, and accessibility of data varies across countries. We seek to create a model of disease spread using data that is accessible, recent, and globally maintained: Roads.
We formulate here a neural network model that can use road layout and density as a surrogate for regional population and connectivity. In learning these connections, it can predict dates of first infection for a given city, and the growth rates thereafter.
The success of our model would inform the optimal distribution of resources, aid and equipment, and could help anticipate the degree and time to enact distancing rules at city scale across the globe.