Several questions regarding Carrot Mapper

[boverdui]

Hello,

I have installed Carrot Mapper locally, to see whether this tool could be useful to us in the future. I have created with WhiteRabbit a scan report of a very simple dataset, which consist of just three tables, i.e. demographic, lab_test, and visit. I have attached the scan report and CSVs to this post. It's all fake data, so nothing that can't be shared publicly.

I have several questions:

(1) How do I specify which field in the source data maps to which field in the OMOP CDM? It looks like Carrot does that automatically, but when I look at the Map Diagram, there are a lot of mappings missing, and other mappings are wrong. For example, for lab_test, only 3 of the 7 fields in the source data are shown in the Map Diagram. How do I specify that visit_id should be mapped to visit_occurrence, that test_value (and not test_type) should be mapped to value_as_number etc. etc.?

(2) I added the concept concept 8532 for the Value F for the Field sex for the Table demographic. This concept now also shows up for the Field test_id in the Table lab_test. I have no idea why. When I remove it there, it also has gone again from the demographic table again.

(3) I tried to add the concept 8845 for the Value pg/ml for the Field test_unit for the Table lab_test. This concept (from the UCUM vocabulary) is present in the CONCEPT.csv in vocabs, but when I try to add it, I get "Adding concept failed. The destination table could not be found or has not been implemented.".

(4) I uploaded a data dictionary, but I don't see any of it's content displayed in the Carrot UI. Shouldn't what is in the "value" column in the data dictionary be displayed in the "Value Description" field in the UI?

(5) Is it correct that it is only possible to archive a dataset, but not to delete it?

With kind regards,
Bert Overduin

ScanReport.xlsx
demographic.csv
lab_test.csv
visit.csv
data_dictionary.csv

[AndrewThien]

Dear Bert,

Thanks for testing and using Carrot Mapper. Also, Sorry for the late reply.

I can answer your questions as below:

1. The Mapping rules saving mechanism of Carrot is described here: https://carrot.ac.uk/mapper/dev_guide/mapping_rules. Also, I don't think the Map Diagram is 100% correct at the moment, yes, and we are planning to archive/improve it in the near future. The correct mappings can always be checked on the Tab Rules though. For example, as the picture below.

This is another topic, maybe, but for the Measurement domain, the source concept id rule will be mapped to "value_as_number". For the "Observation" domain, the source concept id rule will be mapped to "value_as_number" if the column type is number-related, to "value_as_string", if the column type is string-related. But in general, again, the concept will be mapped according to this description: https://carrot.ac.uk/mapper/dev_guide/mapping_rules.

2. I tested using your files and your procedure: "concept 8532 for the Value F for the Field sex for the Table demographic" and everything went well as expected on my end. I am not sure what caused the problems on your end, but maybe you could try re-installing Carrot Mapper? If it happens again, please let me know.

3. Yes, currently all of the concepts having the domain ID: Unit is not supported yet. This domain is not our priority from now on our end. We may discuss more about this and its urgency in our future meetings.

4. Yes, it should display the "value" in the data dictionary as "value description" in the UI. The problem with your data dictionary is the csv_file_name there didn't include the file extension .csv (for example, you put lab_test instead of lab_test.csv). This made Carrot couldn't match the values and their description correctly. I tried adding the csv file extension to your data dictionary and it is working fine now, at least on my end. Please give it a go.

data_dictionary.csv

5. Yes, you cannot delete a Dataset, only Archive it, as a user. Admin of Carrot can delete the dataset through the Django Admin portal though.

I hope this helps with some of your questions and concerns.

Please let me know if you have any other questions.

Thanks!

Thien

[boverdui]

Dear Thien,

Thanks for taking the time to answer all my questions! Although I do like Carrot from what I've seen, it looks like it is not (yet?) useful for us, as a lot of mappings seem not to be supported currently. I appreciate it was originally developed specifically for CO-CONNECT and thus focused on the data from that project. Is it the aim to develop Carrot further, so it can be used for any OMOP project?

Thanks again,
Bert

[AndrewThien]

Hi Bert,
Currently, I would say Carrot is supporting quite a lot of mappings to OMOP tables (as the slide below + device_exposure). Can you please let me know the table mappings that you are looking for and useful for your purposes? We can discuss and accommodate it, if possible.
And I would also say Carrot is still being developed, not only more tables will be supported but also data provenance and high-level features. Hope my answer helps.

Thanks!
Thien

[boverdui]

Hello Thien,

The first OMOP CDM tables that come to mind are:

• Death (which I see is coming)
• Observation_period,
• Visit_occurrence
• Care_site
• Location

and

• Units

This is based on the OMOP projects that we have done thus far. Of course it could be that future projects have additional requirements ....

Cheers,
Bert

[AndrewThien]

Yeah, I see.
Adding supports for more OMOP tables probably isn't the priority for now for Carrot Mapper.
But we definitely will take a look at this later.
Thanks,
Thien

[boverdui]

Thanks, Thien.

Cheers,
Bert