2021-09-20 SEC Call notes

5 min

Opening

USEFUL Jira RESOURCE: CRs I have not accepted.

Berlin Oct 13th SEC meeting - agenda topics

Likely presential:

• IM, PB, TB, ?SI, SG, DB

Remote:

• Pablo, Erik, Shinji

Location

All

Potential items:

• Conformance approach

• Progress on flat JSON / simplified data formats (draft spec)

• REST API - directions / plans ; review current CRs - headers question (@Ian McNicoll , @Erik Sundvall )

• ‘Ians stuff’: AQL link following, terminology tricks, …

• Moving openEHR to ADL2 technology

  • CKM issues; other tools;

  • OPT handling

  • establish scope on value set handling; nested value value sets; asserting relationships;

• next steps AQL: @Seref Arikan : one vendors group AQL call prior to SEC meeting;

  • initial AQL roadmap;

  • work out scope - e.g.demographics; joins; terminology resolution

  • what’s in 2.0? Formalised semantics; what breaks?

  • @Seref Arikan blog posts; permutations

  • IDEALLY: get vendor statements of what permutation handling looks like today (e.g sets v bags)

• ?view Entries & managed lists

• OMOP - @Diego Bosca ? (otherwise on a dedicated call)

15 min

Archie update.

@Pieter Bos

AQL grammar review

@Thomas Beale

See TB Antlr repo.

Discourse discussion.

20 min

@Ian McNicoll AQL - ELEMENT-level querying

Example:
SELECT
el/name/defining_code as someElement
FROM
EHR e
CONTAINS COMPOSITION c
CONTAINS OBSERVATION o
CONTAINS ELEMENT el
WHERE
el/name/value/mappings/target = 'SNOMED-CT::12345'

• could we do ELEMENT-level querying on e.g. OBS-based Entries

• ex: get PEWS total score value via code in defining_code or in mappings

• @Erik Sundvall - definitely exists where need to query on code system A but data are recorded in internal codes, or other code system (hence looking in mappings).

• Assumption: the codes we want to query on are in the data

• Requirements:

  • PRIMARY NEED: get back (e.g) OBS on the basis of term codes used to tag them, without knowing archetype or at-code etc

  • possible subsumption inferencing needed (actual codes in data are children of query code)

  • possible lookup of mappings that are not in the data, but may be in archetypes

  • make the AQL query simpler / more intuitive,

  • dirty import data scenario: need to use the codes that came in the data

  • handle terminology-tagged OBSERVATIONs (presumably other ENTRYs also)

• @Seref Arikan can get a (probably unwanted) permutations effect - question on DISTINCT semantics etc

• ACTION: find out from vendors if ELEMENT-level querying allowed / works / …

Terminology ids & codes problem

@Pablo Pazos

We should agree terminology_id is either

• normal name, decide allowed chars - Unicode sub-ranges; can’t have ‘:' - must be quoted; allow '.’, typical language characters

• URI

See FHIR terminology URI registry

Check also

• Archetype ids - agree that these are [a-zA-Z0-9] based;

• pre-ADL2 - template ids - also agree [a-zA-Z0-9] based

Concept identifier - @Ian McNicoll - potential case for adding an archetype-wide (usually template-wide) id e.g. ‘Scottish BP’. Need is for a human-friendly description - not in data; avoid having to have in technical id

?Optional field; could default to root-node.

NB: Better tools allow Swedish chars in template ids.

Next SEC call

Agenda items for next call:

• @Seref Arikan Archie-based TS generation based on JSON schema.

Outstanding topics: