Open MartinWahnschaffe opened 5 years ago
@danieltomaba @bernardsilenou We need the requirements for every disease that would trigger an outbreak warning or a notification that an outbreak has come to an end.
@danieltomaba Please attach the file you sent me including the answers to the questions I still had to this issue.
@MateStrysewskeSym Disease_Outbreak_warning_def_dta_20190807.xlsx whocdscsrisr992.pdf
We will send you an updated file once my colleagues look at it.
We discussed about the automatic signal detection in SORMAS and decided to use the more sophisticated algorithms for signal detection rather that setting a predefined threshold for each disease. We will begin with the simple methods like the CDC method, RKI methods and latter Farincton methods. There is a package in R called surveillance that has function for these different methods. A first suggestion in implementation is to call R through Java every night to run on the sormas database, apply these algorithms, detect signals and report the output as tables for example. More follows during the next sprint.
@MartinWahnschaffeSymeda @MateStrysewskeSym We need to follow up on this during the next sprint to define how we can integrate the R package via the R server and SORMAS server
@bernardsilenou
@MartinWahnschaffeSymeda @MateStrysewskeSym More information we might need that you might want to have a look: http://www.rforge.net/JRI/ OpenCDU.pdf
It should by possible to define an outbreak warning threshold for each region and disease. The threshold is a number of cases per population within one epi-week.
When the threshold is reached the surveillance supversior of the region should receive a warning notification.
The supervisor can also change the threshold.
@hzi-braunschweig Is surveillance supervisor correct? Can you give an example for a threshold that would be used within a typical region? E.g. 5 cases per 100.000 people? Or better 50 cases per 1.000.000 people?
// based on discussion with Daniel on 21/11/18