EpiInvert is a method to compute Rt and a restored daily incidence curve for the COVID-19. In this release we improve the management of the festive days and allow smooth 7-day variations in the weekly periodic bias correction.

For any questions please contact Luis Alvarez (lalvarez@ulpgc.es).

1. Quick EpiInvert execution for the USA

   • Step 1: Download and unzip the EpiInvert documentation : (WINDOWS VERSION) (LINUX FEDORA RELEASE 11 (in Linux, a recompilation of the source code could be needed. See C++ code below) ).

   • Step 2: Download in the main EpiInvert documentation repository the file owid-covid-data.csv (see reference [4] below)

   • Step 3: Execute the batch file EpiInvertBatch.bat ( double-clicking in Windows). The outcomes of the execution are stored in the files SummaryEpiInvertExecution.txt and Rt.csv

2. Understanding EpiInvert execution outcomes (see [1] for more details)

   • The file SummaryEpiInvertExecution.txt contains basic information about the EpiInvert execution including the weekly correction factors.

   • The file Rt.csv contains the following daily value sequences:

     1. "q seasonality" : correction factors of the incidence weekly periodic bias.

     2. "date" : Date of the daily value.

     3. "festive day" : Declaration of festive days used.

     4. "Rt EpiInvert" : EpiInvert Rt daily estimation.

     5. "Rt EpiEstim" : EpiEstim Rt daily estimation,

     6. "original incidence" : Original daily incidence curve

     7. "festive day bias corrected" : Original daily incidence curve corrected using the festive days.

     8. "weekly bias correction" : Daily incidence curve corrected using the weekly periodic factors.

     9. "renewal equation" : Restored incidence curve using the renewal equation.

     10. "Rt variability" : EpiInvert Rt variability in the last 3 days.

     11. "Radius Rt 95% CI" : Empirical estimation of the radius of a 95% confidence interval for Rt.

     12. "Radius Rt 90% CI" : Empirical estimation of the radius of a 90% confidence interval for Rt.

     13. "Radius incidence 95% CI" : Empirical estimation of the radius of a 95% confidence interval for the restored incidence (in the case of forecasting).

     14. "Radius incidence 90% CI" : Empirical estimation of the radius of a 90% confidence interval for the restored incidence (in the case of forecasting).

3. EpiInvert execution for any country using the data from owid-covid-data.csv.

   • Step 1: Edit the file MyFestiveDays.csv and put the festive days you want to use for the country in the same proposed format (this file can be empty or, equivalently, you can remove the file name from the batch file).

   • Step 2: Edit the file EpiInvertBatch.bat and replace the string USAFestiveDays.csv with MyFestiveDays.csv (or remove the file name) and replace USA with the country acronym in owid-covid-data.csv (for instance DEU for Germany)

   • Step 3: Execute the batch file EpiInvertBatch.bat

4. EpiInvert execution using your own incidence data

   • Step 1: Edit the file MyFestiveDays.csv and put the festive days you want to use for the country in the same proposed format (this file can be empty, or, equivalently, you can remove the file name from the batch file).

   • Step 2: Edit the file MyIncidence.csv and introduce your daily incidence data using the proposed format. If you do not want to include festive days you do not need to add in the data file the date information of each incidence value (you can just remove the row with the date values).

   • Step 3: Edit the file EpiInvertBatch.bat and replace the string USAFestiveDays.csv with MyFestiveDays.csv (or remove the file name) and replace USA (the first parameter) with MyIncidence.csv

   • Step 4: Execute the batch file EpiInvertBatch.bat

5. Understanding EpiInvert execution parameters

The file EpiInvertBatch.bat contains the default EpiInvert command line execution instructions given by "WindowsEpiInvert.exe USA Ma.txt 5 CASE true 8 true USAFestiveDays.csv 150". The meaning of these parameters (that you can modify) are:

• Parameter 1 (default USA) : Country acronym (in the case of taking the incidence from owid-covid-data.csv) or the file name with your daily incidence data. The only mandatory parameter is this one.

• Parameter 2 (default Ma.txt) : File with the serial interval, you can use Ma.txt, Du.txt, Nishiura.txt or any one you want (keeping the proposed format).

• Parameter 3 (default 5) : Rt regularization weight of EpiInvert method (see [1] for more details)

• Parameter 4 (default 5) : Seasonality regularization weight of EpiInvert method (see [1] for more details)

• Parameter 5 (default true) : Boolean value to select if you want to use the weekly bias correction (see [1] for more details).

• Parameter 6 (default 3) : Number of weeks in the past used to perforem a 14-day forecast of the incidence. The values can be 0 (no forecast), 1, 3 or 5.

• Parameter 7 (default 200 ) : The maximum number of days in the past used by EpiInvert. It is used to reduce the computational cost when you are interested just in the more recent dates.

• Parameter 8 : The file containing the festive days you want to use (it can be empty) (see [1] for more details).

6. C++ code

In the repository "C++ code" you can find the EpiInvert source code. By compiling this code in any operating system you can run EpiInvert using the command line as showed in the batch file EpiInvertBatch.bat. In the case of LINUX maybe you will need to recompile the provided precompiled code using the provided makefile.

7. Licence

CC Creative Commons "Attribution-NonCommercial-ShareAlike"

Bibliography:

[1] Luis Alvarez, Jean-David Morel and Jean-Michel Morel (2022). "Modeling COVID-19 Incidence by the Renewal Equation after Removal of Administrative Bias and Noise", Biology 11(4), 2022.

[2] Luis Alvarez, Miguel Colom, Jean-David Morel and Jean-Michel Morel (2021). "Computing the daily reproduction number of COVID-19 by inverting the renewal equation using a variational technique", PNAS, 118 (50), pp. 1-10, 2021.

[3] "Short time forecasting of the COVID-19 incidence curve by learning from incidence evolution in the past", Preprint

[4] Hannah Ritchie, Edouard Mathieu, Lucas Rodés-Guirao, Cameron Appel, Charlie Giattino, Esteban Ortiz-Ospina, Joe Hasell, Bobbie Macdonald, Diana Beltekian and Max Roser (2020) - "Coronavirus Pandemic (COVID-19)". Published online at OurWorldInData.org. Retrieved from: 'https://ourworldindata.org/coronavirus' [Online Resource]