Covid Fox Monitoring
A Covid Wastewater Project
- Give People An Overview of Current Covid Transmission Levels
- Collect & Share Data Sources for people to use for their Covid-projects
- Find more Datalinks to Covid Wastewater Data (Please suggest!) 🔨👷🚧
⚠️ ${\color{darkorange}\text{In Progress}}$ - Find ways to use social media data ❌
${\color{red}\text{Cancelled}}$ - Find ways to pull aggregated google trends search terms data by region
- Find ways to pull aggregated mobility data per region
- Forecast covid transmission per country 1-4+ weeks ahead ✔
${\color{green}\text{Completed}}$ - Forecasts per region
- Improve forecasts
- Add variant data for better predictions ❌
${\color{red}\text{Cancelled}}$ - Improved data standardization ❌
${\color{red}\text{Cancelled}}$ - Add Research News from Nature.com using LLM to interpret Abstract for laypersons.✔
${\color{green}\text{Completed}}$
- Find ways of better hosting (URGENT - Exceeding free github tier) ✔
${\color{green}\text{Completed}}$ - Data Quality Control
- Implementation of proper CI/CD pipeline
- Testing prior to releases
- Full automation of independent daily updates
\mathscr
- Simulations of NPI's & transmission level
Link to Covid-19 Wastewater Monitoring Website (Click Here!)
Interpretation
- Sweden: Red = "High Transmission". Based on the height of Uppsala wastewater measures during wave 1 2020.
- All Other Countries: Red = "Relatively High Transmission". Based on each country separately relative to the Min-Max values in the timeperiod.
- Withinin country comparisons can be made.
- Between country comparisons should ONLY be made with careful interpretation. The colors are relative, so I believe only relative time dimension interpretations are possible. NOT degree of transmission on a specific week. The reason is that almost each country uses different metrics to track transmission.
The Heuristic of the Swedish Cut-off Value of 10 and above indicated as "high transmission" (red). (EDIT 2024-12-11: Equivalent value of 10 in new source file scale is 660 000)
Later might update the trend graphs. The trick is to figure out how to create a concise & intuitive overview.
A simple Neural Net implemented. Caution: Very early stage. Interpret carefully. Known issues:
- Bugs when running for some countries e.g. Finland and Poland.
- Will not capture sudden strong spikes.
Uses a large language model to make research news more accessible to non-scientists.
For questions, suggestions, requests or ideas: Contact me on twitter or linkedin.
LinkedIn