Hi Thomas, yes - it has been discussed quite carefully 
.
A differential diagnosis is usually only a temporary or short term placeholder for one or more possible diagnoses, each of which then gets accepted as a diagnosis fit for transforming to the status of using the EVAL.problem_diagnosis or excluded - hence the states of possible/accepted/excluded.
The Problem/diagnosis qualifier archetype (nested within the EVAL.problem_diagnosis) tries to cater for all the various flavours of diagnoses used in different physical/professions/temporal contexts , including different types of certainty. It is a complex area, extremely variable and inconsistent in data sets. We decided not to add certainty in Differentials, but have added an ‘order’ element which could be used to indicate differentials that are more likely or are higher priority for further investigation.
To be perfectly honest, I’m not aware of differentials being shared, and certainly not with percentages. If we’ve missed something and anyone has use cases, please feel free to share before it is published.
This is why we post these notifications - another mechanism to get input in case we’ve missed inviting a critical reviewer.