How to share PLP results without including patient level data?

Researchers
1

When PLP model is trained, the results will be held in the runPlp.rds and there is one table named prediction that has patient level information. Now when I want to share the prediction results with external collaborators, I want that they would be able run the shiny app, but I’m not allowed to share patient level tables with them. How can I generate shareable results in a way to keep the data sufficiently private? For example, turn off the inclusion of the prediction table in the results.

Just deleting this table from runPlp.rds file broke the viewPlp function.

2

Hi @Raivo_Kolde ,

There is a function called savePlpShareable which should save it in a form without patient level data. Then this can be loaded with loadPlpShareable and viewed with viewPlp . Try the following snippet to test it:

library(PatientLevelPrediction)
data("simulationProfile")
plpData <- simulatePlpData(simulationProfile, n = 1000, seed = 42)
saveLoc <- file.path(tempdir(), "savePlpShareable")
results <- runPlp(plpData, saveDirectory = saveLoc)
newSaveLoc <- file.path(tempdir(), "savePlpShareable", "saved")
path <- savePlpShareable(results, newSaveLoc)
shareableRunPlp <- loadPlpShareable(path)
viewPlp(shareableRunPlp)

This at least opens in the shiny and seems to work for me. Let know if it doesn’t for you.
Regards,
Egill