Deidentification

[jeremyestein]

The following data items are considered sufficiently identifying that we need to replace them with non-identifying values before we send them to the DSH:

• MRN
• CSN
• bed location

(anything else?)

I'm assuming that we are not pulling in names, DoBs, etc in the first place.

Sub-tasks:

• create a psuedo anonymous version of the controller output.
• check that dev keyvault works
• Talk to Dave about how the prod keyvault credentials will work
• implement real hashing through the hasher
• test it with dev keyvault using synthetic data (not on hospital systems)

[jeremyestein]

There is a toy hasher for the sake of getting the pipeling running. This will need to be replace with the real hasher as part of resolving this issue.