FHIR Store: the FHIR standard in your favourite format

FHIR Store provides the FHIR standard in several formats, including json and yml.

yml standard

We have converted the FHIR specification available on this hl7 page (See R2 Layout tab) into the yml format.

The folders follows the hierarchy described in the R2 Layout, and an additional folder DataTypes retrieves the complex types detailed on this hl7 page

Note that the json format is also available, which can be more easy for api calls. Oh, actually we already built an API just here!

Typing convention

The FHIR standard has a structure of a tree. We have typed each node, which can be a simple type (eg deceasedBoolean<boolean> or birthDate<date>). One basic type which is worth explaining is code, which is used like this: gender<code=male|female|other|unknown>. If no type is provided, we assume it is a string. You can also have special DataType (eg HumanName, ContactPoint, Address, see complete list) which follows the same convention (eg address<Address>). If the node is a list, we add list:: in the typing, like in telecom<list::ContactPoint>. Note that for the case of a list of strings we will have node<list>.

Bindings conventions

You have 4 options to link a FHIR attribute to SQL resources, which can be combined. Have a look in the fhir-mapping project at Identification/Individuals/Patient.yml to see these examples used in a real context.

1. _col: the name of the column where the information is visible. You should specify <OWNER>.<TABLE>.<COLUMN>. You can put several items in a list if the information is dispatched.
2. _script: the name of the script that you apply on the _col items to clean the data. For phone it is clean_phone, etc; if none just put equal. Usually you have 1 script for 1 col, but in the case where you have 1 script and n>1 cols, then all the cols are given as arguments to the script.
3. _join: if the information is not available in the "root" table (eg ICSF.PATIENT for Resource Patient), you specify the join to do on another table to get the info. Join has two attributes,
   • _type which specifies the nature of the join (OneToOne, OneToMany, etc) and
   • _args which details how to join with the format ["<owner>.<table>.<col>=<owner>.<join_table>.<join_col>"].
4. _template_id: when the node is of a special DataType (eg HumanName, ContactPoint, Address, etc), the description of each subfile will be made in the yml file which has the same name, in a template referenced by an ID. This helps having a short Resource mapping file.

How to start

We have reported several issues with the label Good first issue which can be a good way to start! Also of course, feel free to contact us on Slack in you have trouble with the project.

If you're enthusiastic about our project, :star: it to show your support! :heart:

License

Apache License 2.0