A very broad question I had asked but my set up like this. So I took a data that was published in 2013-14 when nejm paper for AML came out, so the criteria i took is FAB classification then , Cytogenetic risk and other clinical factor. Then another study published in 2018-19 called BeatAML which had again same criteria but the FAB classification were bit more refined where as in 2013-14 it was bit primitive sort of.
So my initial model was with the old data that was my test model and the new one is for my validation cohort to predict patient survival. Now i do see if i follow new classification my old analysis will be modified a bit as samples with be again reassigned to one or the other group which i don't want to do as my paper is in final stage.
Now how do i justify or what would be the error with the approach or potential question that might come.
I would like to hear from stats people who can guide me.
Any suggestion or help will be really appreciated