Incorporate CDE definitions from RadElement as `FindingModel` objects

[talkasab]

• Script that converts JSON in the CDE format (v1.0) to a FindingModel
  • Model IDs should look like, OIFM_CDE_000195 (for RDES195)
  • Attribute IDs should look like, OIFMA_CDE_001300 (for RDE1300)
  • Value IDs should follow the pattern
  • At each level, incorporate a code reference to the original CDE:

    {
      // ...
      "index_codes": [
        {
          "system": "RADELEMENT",
          "code": "RDES195",
          "display": "pulmonary nodule"
        }
      ]
      // ...
    }

  • If there's a body_parts defined at the top level, copy it over to the index_codes at the top level
  • Dealing with the values in value sets:
    • Map definition to description in general
    • If definition is just the same text as name, you can skip it (even if capitalized)
    • The name in FM should come from either value or name in the CDE
    • If both name and value are defined in CDE, put the CDE value in the FM name and put the CDE name into the FM description field (combine it with the CDE definition separated by a : if definition is also defined)
    • If the values are just (1, 2, 3, 4, ...) (as in RDES126), just skip the values and use CDE name as the name in the FM
• Get the CDE definitions (*.cde.json) from the OIDM CDE repository and convert them all
• Create a new branch on findingmodels and put the new *.fm.json files in the defs directory.

[talkasab]

Cleaning Up Choice Elements

Now we know that name always has a good value, and value is junk. So the steps for converting the values of a choice element are:

• name → name
• definition → description if the description is not the same as name (casefolded), otherwise skip it
• value_code should be OIFMA_CDE_000111.0, where the number after the digit is the zero-based index of the value

[talkasab]

Example CDE → FM JSON:

CDE

    {
      "id": "RDE818",
      "parent_set": "RDES126",
      "name": "Aortic dissection",
      "definition": "",
      "question": "",
      "schema_version": "1.0.0",
      "element_version": {
        "number": 1,
        "date": "2023-10-05"
      },
      "status": {
        "name": "Published",
        "date": "2023-10-05"
      },
      "index_codes": [],
      "contributors": {
        "people": [],
        "organizations": []
      },
      "history": [],
      "references": [],
      "value_set": {
        "min_cardinality": 1,
        "max_cardinality": 1,
        "values": [
          {
            "code": "RDE126.0",
            "value": "0",
            "name": "acute",
            "definition": "Acute"
          },
          {
            "code": "RDE126.1",
            "value": "1",
            "name": "subacute",
            "definition": "Subacute"
          },
          {
            "code": "RDE126.2",
            "value": "2",
            "name": "chronic",
            "definition": "Chronic"
          },
          {
            "code": "RDE126.3",
            "value": "3",
            "name": "indeterminate"
          },
          {
            "code": "RDE126.4",
            "value": "4",
            "name": "unknown"
          }
        ]
      }
    },

FM

{
  "oifm_id": "OIFM_CDE_000126",
  "name": "Acute Aortic Syndrome",
  "description": "This module describes the Common Data elements and Macros for Acute Aortic Syndrome",
  "attributes": [
    {
      "oifma_id": "OIFMA_CDE_000818",
      "name": "Aortic Dissection",
      "description": "Aortic dissection",
      "type": "choice",
      "values": [
        {
          "value_code": "OIFMA_CDE_000818.0",
          "name": "acute"
        },
        {
          "value_code": "OIFMA_CDE_000818.1",
          "name": "subchronic"
        },
        {
          "value_code": "OIFMA_CDE_000818.2",
          "name": "chronic"
        },
        {
          "value_code": "OIFMA_CDE_000818.3",
          "name": "indeterminate"
        },
        {
          "value_code": "OIFMA_CDE_000818.4",
          "name": "unknown"
        }
      ]
    }
  ],
  "index_codes": [
    {
      "system": "RADELEMENT",
      "code": "RDES126",
      "display": "acute aortic syndrome"
    },
    {
      "system": "RADLEX",
      "code": "RID480",
      "display": "aorta"
    }
  ]
}

[talkasab]

Writing the FM JSON files

• Make sure to use model_dump_json(indent=2,exclude_none=True)
• Use the function findingmodel.common.model_file_name on the name field to create the appropriate name to write to

from pathlib import Path

from findingmodel import FindingModelFull
from findingmodel.common import model_file_name

fm: FindingModelFull = ... # Complex process of generating a new FM from CDE
path = Path(model_file_name(fm.name)).write_text(fm.model_dump_json(indent=2,exclude_none=True))