New forecast: by Aug 1 Florida will report 130-290 #covid19 deaths per day (7-day moving avg)
This is based on applying age-stratified CFR adjusted for right censoring to the state's line list.
My model's open source:
github.com/mbevand/florid… & has great track record...
1/N
This is based on applying age-stratified CFR adjusted for right censoring to the state's line list.
My model's open source:
github.com/mbevand/florid… & has great track record...
1/N
My first forecast () accurately predicted the recent significant increase of deaths in Florida. The red curve below shows deaths reported AFTER I published my forecast:
2/N
2/N
List of the previous forecasts:
Forecast as of 7/5:
Forecast as of 7/11:
Forecast as of 7/14:
3/N
Forecast as of 7/5:
Forecast as of 7/11:
Forecast as of 7/14:
3/N
Details: all is documented in the GitHub repo: github.com/mbevand/florid…
Essentially I use the state's line list to calculate the Gamma distribution of time between onset of symptoms and death. I use it to calculate the age-stratified Case Fatality Ratio adjusted for censoring
4/N
Essentially I use the state's line list to calculate the Gamma distribution of time between onset of symptoms and death. I use it to calculate the age-stratified Case Fatality Ratio adjusted for censoring
4/N
I calculate a mean onset-to-death time of 18 days.
Based on the state line list that documents the date of onset of symptoms, we can thus accurately predict deaths up to ~18 days ahead of time.
Again, for all the details see github.com/mbevand/florid…
5/N
Based on the state line list that documents the date of onset of symptoms, we can thus accurately predict deaths up to ~18 days ahead of time.
Again, for all the details see github.com/mbevand/florid…
5/N
