Inspiration
Our idea is to provide a better insight on the status of the disease on a level of cities and counties. Often the spread of the virus varies in different areas and we want to give a better insight to the local decision makers of how much their county is at risk of rising infection rates. We therefore can better implement precautionary actions such as lockdowns or orders of ventilators. We can see that the restrictions can vary from region to region and since the Corona pandemic has just started this tool shall give great power for the upcoming months, when we will see new waves of Corona infections in different areas in Germany.
What it does
We use the dataset provided by the Robert-Koch Institut and added new features, such as population density, demographic information, employment rates, jobs by industry etc. Moreover we want to use geodesic information, weather data and Epidemic models, e.g. the SEIR model like it was used by Wang et al.
How we built it
We do a lot of preprocessing on the data to get a clean dataset that we can work with. Moreover we started at the same time with a visualisation of Germany and its counties. We want to implement different visualisation techniques such as color coded counties indicating the cases per 100.000 inhabitants that may change based on the predictions of our models. The user may change a slider and can see how the different counties evolve based on our forecast. We can additionally provide ticking boxes that correspond to different political restrictions or behaviour of the people. With that tool in hand we can give a small forecast on a county level and can early on detect counties that may be at risk of exploding infection cases.
Challenges we ran into
The main challenge is to make a robust prediction. One key element is that we need data that can provide enough information that has a significant effect on the spread of the virus.
Accomplishments that we're proud of
We started building up a dataset that can still be very handful for different analysis in the future. Since the Corona Pandemic will be still around for at least a year, we are proud to provide a first database that can be used for other future data analysis tasks around the Corona Pandemic. We think that is crucial to build up a better centralised database around the Corona Crisis. There is a lot of data gathered at the moment, but it still is Hardtop access raw data.
What we learned
Predicting the spread of a disease on such a small level is very difficult and can have so many causes that are often hardy describable by sociological, demographical or infrastructural data.
What's next for Corona Cases Forecasting for Germany on a County Level
- Set up an API that fetches the new data on a daily basis
- Implement a epidemiological model such as SEIR.
- Collect more data such as unemployment rates, weather data, historic flu epidemic data.
- build a website that visualises the data on a map.
References
Wang, H., Wang, Z., Dong, Y. et al. Phase-adjusted estimation of the number of Coronavirus Disease 2019 cases in Wuhan, China. Cell Discov 6, 10 (2020). https://doi.org/10.1038/s41421-020-0148-0
Log in or sign up for Devpost to join the conversation.