201605 Declarative Mapping
Return to May 2016 Proposals - Return to FHIR Connectathon 12
Declarative Mapping
This track will explore mappings and/or a useful subset of mapping tasks to support interoperability between FHIR and other data representations. The intent is to enable mappings to be declared, transmitted and executed to translate data from one form to another (particularly FHIR). Mappings could be written in any style, ranging from simple translation tables or expressions to executable code such as EcmaScript. Different participants may demonstrate the virtues of different mapping styles.
- Pre-requisites: none
Schedule
Saturday 1pm-4pm:
- Participants show & tell
- Draft summary and outcomes
Sunday 9am-12:30pm:
- Finish up summary and outcomes -- See report
- Start planning for Baltimore FHIR Connectathon
Submitting WG/Project/Implementer Group
- SOA Working Group
- CGIT (approval planned)
- CBCC (approval planned)
Justification
There is a recognized need for mappings to facilitate interoperability across HL7 products, and between HL7 products and other formats and data models. Following on the Birds of a Feather CDA-to-FHIR session in Orlando, there was an agreement among the participants to pursue declarative mapping approaches. Objectives of the track include developing and demonstrating reusable computable artifacts that can be saved, persisted and transmitted as FHIR resources. The SOA WG has already developed "cross-paradigm" mapping guidance intended to represent the same business content in more than one structure (e.g. CDA template to FHIR profile or V2 profile components).
See also http://www.healthintersections.com.au/?p=2506
Proposed Track Lead
Expected participants
The expected participants include:
- Keith Dudley / Grahame Grieve
- MDMI Open Source Project (several participants)
- David Booth (Yosemite Project)
- SAMHSA BHITS Project
- Members of the i2b2 on FHIR team
- David Hay (Orion Health)
Roles
Map Designer
This is a software product that helps a user manage a set of mapping resources, and edit them, and test the mapping transform outcome interactively. Designer may offer additional assistance in Ux for mapping etc as desired. Designers implement use case #1.
Map Executor
This will be an application that takes a set of map files or resources, a CDA document and produces a FHIR Bundle Document. The executor may be wrapped in any form desired.
Scenarios
TODO: David Booth and Ken Lord to add more use case scenarios, with before and after documents.
1. Edit an existing map
Action:
The designer opens and edits the existing CDA mappings either from the map files (using the concrete syntax), or from the equivalent resources.
Success Criteria:
The map resources produced can feed into the second use case
Bonus Point:
Bonus points for being able to read and write resources from a server.
2. Execute a map
Action:
Take the CCDA draft maps, apply them to a CCDA example document, and produce a document bundle
- The test map is the one in the CCDA implementation (\maps)
- the test document is the CCDA example document distributed with the CCDA R2 (see [here]).
Success Criteria:
The output should be a Bundle that looks like [this - to be resolved closer to the day]
Future Step:
A future step will test the execution packaged in an operation on a FHIR Server.
3. Translate data (Yosemite Project / cMUMPS data)
Start condition: A patient record extracted from an unspecified cMUMPS-based system as JSON-LD, containing demographics, prescriptions and lab results. (cMUMPS is used here as in lieu of an existing MUMPS-based DOD electronic health record system that shall remain nameless.)
End condition: A FHIR RDF document containing representative source data translated into the FHIR resources.
Process: Using sharable, pluggable translators, translate source data to target data by running each translator (on an appropriate translation engine) and automatically assembling the results.
Extra Credit: Demonstrate how translators written in different languages can be used together.
4. Translate data (Yosemite Project / AALTER data)
Start condition: Data describing multiple procedures performed on a single patient, extracted from an AALTER system as JSON-LD. (AALTER is used here in lieu of an existing DOD electronic health record system that shall remain nameless.)
End condition: A FHIR RDF document containing data expressed as FHIR Procedure resources.
Process: Using sharable, pluggable translators, translate source data to target data by running each translator (on an appropriate translation engine) and automatically assembling the results.
Extra Credit: Demonstrate the use of sharable, pluggable translators to produce a FHIR Bundle in FHIR RDF, containing both the results of this use case and Use Case #3.
