Closed ksdkalluri closed 5 years ago
is this change intentional?
Let me clarify the matter. Initially, all source concepts are considered to be Non-standard. Only after the mapping their domains can be defined. Being mapped to Standard concept source concept inherit its domain. Not being mapped to a Standard concept a source concept becomes standard and get a domain whichever you prefer.
The old PPI mapping logic (whose results we are seeing in Athena now) is the following: 1) All PPI source concepts have one-to-one mapping. 2) If PPI source concepts have standard equivalents in SNOMED or LOINC, they are considered to be Non-standard 3) If PPI source concepts do not have standard equivalents in SNOMED or LOINC, they are considered to be Standard 4) The domains used for Questions are 'Measurement' and 'Observation' (depending on the existence of a mapping to a Standard concept). 5) The domains used for an Answers vary. They can be represented by Observation, Meas Value, Condition or Procedure.
Due to regular PPI mapping refreshes (we have already performed 6 ones), domains of PPI vocabulary was being changed respectively .
According to the 'valid_start_date' field in the CONCEPT_RELATIONSHIP table, domains of PPI concepts you mentioned were assigned either in 2017-09-12 (the second version of the mapping) or in 2018-08-15 (the fifth one). That was fine in respect of the old PPI mapping logic. But the times have changed.
Currently we are working on the implementation of the new PPI mapping logic (see the diagram below). Long story short, all PPI source concepts will get 'Observation' domain to be easily connected with the SURVEY_CONDUCT table. To aim this, we are also changing the domain distribution of the LOINC vocabulary which is widely used in mappings of any survey-related entities as well as PPI Questions.
So very soon you will see improved structure of PPI vocabulary, which can be used as a target vocabulary for mappings of other questionnaires to standard concepts in the future (in the PPI, the are currently 1237 standard concepts).
we also found that 139 PPI to PPI concept_ancestor records missing in the new version, Not sure if it both are related.
The hierarchy was also being modified due to the customer requests. Could you provide us with the list of these PPI concepts to let us confirm the relevance of such changes?
This is very helpful, Thank you.
All of Us Research Program has been using v5.0 22-DEC-17 and we are planning to upgrade to v5.0 17-MAR-19. In Concept table there are a lot of concepts for which the domain has been changed since the old version for the PPI Vocabulary. for eg.
gives result
gives result
To make sure I checked http://athena.ohdsi.org for one of the concept_id from the above query 1585260 which is concept_id for e-mail. I found out that the
Non-standard to Standard map(OMOP)
is linked toLOINC
instead ofPPI
.For all the concepts with
Non-standard to Standard map(OMOP)
linked toLOINC
Vocabulary instead ofPPI
, the domain is being changed fromobservation
tomeasurement
.But most of the concepts like e-mail, First_name, phone etc. belong to
Observation
domain rather that themeasurement
.is this change intentional?
we also found that 139 PPI to PPI concept_ancestor records missing in the new version, Not sure if it both are related.