Whitepaper on Document Sharing Across Network Topologies expands upon the concepts in the Health Information Exchange White Paper by providing additional guidance on how existing document sharing communities can be interconnected to form a unified federated exchange ecosystem.
Grouping of actors is mentioned in section 1:46.8.3.
Does Karen's Cross apply here? If so, how? Should OrganizationAffiliation be required?
Karen's Cross
(see 3.0 Interaction Patterns. Also described here)
is a mapping table defined by the Direct Project (not by IHE),
that tells how to get to and from different flavors of IHE Document Sharing "Push" (XDR, XDM) and the Direct Protocol.
It was done at a "whiteboard" level of detail, and resulted in specific requirements for transforming
messages more or less isomorphically from one flavor to another. Later, additional requirements
were added for encoding Direct addresses in XDR SubmissionSet.intendedRecipient. It should be noted
that the Cross is incomplete; neither Direct nor IHE has any analogous requirements for transforming,
say, an XCA Query and Retrieve response into an XDM file. XCDR and MHD Push and Pull are also missing.
That said, IHE Document Sharing profiles (not counting Direct) are generally considered similar enough that
transformations should be obvious.
So, when would Karen's Cross (or an expanded version) apply? Potentially anywhere two systems
need to translate between different Doc Sharing actors, but it's really only needed if transformations
are not obvious. Maybe it needs to be documented by IHE (especially if it's between IHE actors).
We'll look into this in a second.
But first, OrganizationAffiliation shouldn't be required, because it's orthogonal
to whether translation is needed. For example: In the directory, Org A has two Endpoints:
an MHD Document Responder and an XDS Document
Registry/Repository. Behind the scenes, the MHD actor is an adapter over the XDS actors.
This is simply two APIs to the same organization. The HCID and any other
organization or author identifiers are simply copied; there is no translation needed.
Now let's consider the other cases where there's federation to other organizations/entities
not directly reachable. OrganizationAffiliation is just one case:
An Org allows access to others related with partOf
An Org allows access to others related with OrganizationAffiliation
An Org allows access to Practitioners related via mCSD links
In these cases, there might be a translation layer behind the exposed Endpoint to get to those
other entities, or there might be some other proprietary mechanism: internal EHR messaging,
direct DB access, etc.
So would Karen's Cross (or an expanded version) potentially apply? Yes, but likely only in making
sure that addressing of federated organizations/entities is clear. We have that as issue mCSD_18.
Grouping of actors is mentioned in section 1:46.8.3. Does Karen's Cross apply here? If so, how? Should OrganizationAffiliation be required?
Karen's Cross (see 3.0 Interaction Patterns. Also described here) is a mapping table defined by the Direct Project (not by IHE), that tells how to get to and from different flavors of IHE Document Sharing "Push" (XDR, XDM) and the Direct Protocol. It was done at a "whiteboard" level of detail, and resulted in specific requirements for transforming messages more or less isomorphically from one flavor to another. Later, additional requirements were added for encoding Direct addresses in XDR SubmissionSet.intendedRecipient. It should be noted that the Cross is incomplete; neither Direct nor IHE has any analogous requirements for transforming, say, an XCA Query and Retrieve response into an XDM file. XCDR and MHD Push and Pull are also missing. That said, IHE Document Sharing profiles (not counting Direct) are generally considered similar enough that transformations should be obvious.
So, when would Karen's Cross (or an expanded version) apply? Potentially anywhere two systems need to translate between different Doc Sharing actors, but it's really only needed if transformations are not obvious. Maybe it needs to be documented by IHE (especially if it's between IHE actors). We'll look into this in a second.
But first, OrganizationAffiliation shouldn't be required, because it's orthogonal to whether translation is needed. For example: In the directory, Org A has two Endpoints: an MHD Document Responder and an XDS Document Registry/Repository. Behind the scenes, the MHD actor is an adapter over the XDS actors. This is simply two APIs to the same organization. The HCID and any other organization or author identifiers are simply copied; there is no translation needed.
Now let's consider the other cases where there's federation to other organizations/entities not directly reachable. OrganizationAffiliation is just one case:
In these cases, there might be a translation layer behind the exposed Endpoint to get to those other entities, or there might be some other proprietary mechanism: internal EHR messaging, direct DB access, etc.
So would Karen's Cross (or an expanded version) potentially apply? Yes, but likely only in making sure that addressing of federated organizations/entities is clear. We have that as issue mCSD_18.