Problems with Encounter and Oberservations

[Goldkugel]

Dear Projectathon team,

we are trying to run the docker containers and are struggling with two issues:

1. Our encounters on the fhir server do not have any "type"-field. Therefore, if one filters by this field (in the script it filters by type="Einrichtungskontakt") the result will be always empty. Is there any possibility to solve this e.g. not filtering at all?

2. Our observations have as a patient resource an identifier and not a URL. It seems that the script has a problem when it faces an observation which has an identifier as a resource. In this case, the resource object is not being set. Are we correct? Regarding this issue, we attached a screenshot which might be helpful to understand.
   

Thank you for your help in advance.

Best regards

AIIM Team of TUM Munich

[juliangruendner]

@Goldkugel:

1. if the type "Einrichtungskontakt" is not set - it is not implemented correctly.
   However you could adjust the data_extraction_config.json accordingly

2. We expect a reference to be set - if u are currently not setting a reference u should adjust your ETL jobs accordingly.

[RaffaelBild]

@juliangruendner Thank you very much for the fast reply!

Regarding point 1, I would like to add that the attribute type has the cardinality 0..*
(see https://simplifier.net/medizininformatikinitiative-modulfall/kontaktgesundheitseinrichtung)
and hence, it might be more robust to restrict the FHIR search query in the script to MII Encounter resources using the attribute Encounter.meta.profile rather than Encounter.type.

Regarding point 2, I would like to point out that the profile MII-Reference supports both references via the attributes reference and identifier (see https://simplifier.net/packages/de.medizininformatikinitiative.kerndatensatz.meta/1.0.3/files/319433).
In our ETL jobs, we set the Observation.subject.reference to the URL of the corresponding Patient resource whenever it has already been materialized on the FHIR Server. However, if not, we set the Observation.subject.identifier to a business identifier of the Patient that has yet to be loaded. I think both should be in accordance with the MII profile and it would be perfect if the script would just ignore those Observation resources that do not have Observation.subject.reference set. Is it feasible to implement this? That would be much appreciated.

Best regards
Raffael

[juliangruendner]

@RaffaelBild:

1. as mentioned above you can remove the encounter type from the query in the config:
   https://github.com/juliangruendner/pj7-data-extraction/blob/main/data_extraction_config.json.default#L24

2. We will not support the reference by identifier for this projectathon

[Goldkugel]

Dear Julian,

we do not need support for references by identifier. We would kindly ask you to modify slightly your script in such a way that references by identifiers are ignored. Would that be suitable?

Thank you for your help.

Best

AIIM Team of TUM Munich

[juliangruendner]

@Goldkugel - what do you mean by ignored - deleted from the resource in the pseudonymisation stage?

[Goldkugel]

It would be great if Observations which have identifiers as references are not processed any further and are not part of the result or any calculation in it.

[juliangruendner]

@Goldkugel - only observations or any resources?

[Goldkugel]

Only observations.

[juliangruendner]

i think this would be a change very specific to your site.
I would recommend that you fork the repo and create your own version.

[juliangruendner]

@Goldkugel - also be aware that this repo is deprecated:
for new version please see: https://github.com/medizininformatik-initiative/Projectathon7-VHF-DataExtraction