Feature/130 cohort ontology

tracex_project/extraction/logic/modules/module_cohort_tagger.py

@@ -36,6 +36,7 @@ def execute_and_save(
 
         cohort_dict = self.__extract_cohort_tags(patient_journey)
         cohort_dict = self.__remove_placeholder(cohort_dict)
+        cohort_dict = self.normalize_coniditons_snomed(cohort_dict)
 
         return cohort_dict
 
@@ -65,3 +66,23 @@ def __remove_placeholder(cohort_data) -> Optional[Dict[str, str]]:
             return None
 
         return cohort_dict
+
+    @staticmethod
+    def normalize_coniditons_snomed(cohort_dict) -> Optional[Dict[str, str]]:
+        """Normalizes conditions to a SNOMED code."""
+        condition = cohort_dict.get("condition")
+        preexisting_condition = cohort_dict.get("preexisting_condition")
+
+        if condition is not None:
+            (
+                cohort_dict["condition"],
+                cohort_dict["condition_snomed_code"],
+            ) = u.get_snomed_ct_info(condition)
+
+        if preexisting_condition is not None:
+            (
+                cohort_dict["preexisting_condition"],
+                cohort_dict["preexisting_condition_snomed_code"],
+            ) = u.get_snomed_ct_info(preexisting_condition)
+
+        return cohort_dict

tracex_project/extraction/migrations/0022_add_snomed_trace_cohort.py

@@ -0,0 +1,53 @@
+# Generated by Django 4.2.13 on 2024-05-26 09:35
+
+from django.db import migrations, models
+import django.db.models.deletion
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('extraction', '0021_alter_event_event_type_alter_event_location_and_more'),
+    ]
+
+    operations = [
+        migrations.RemoveField(
+            model_name='trace',
+            name='cohort',
+        ),
+        migrations.AddField(
+            model_name='cohort',
+            name='condition_snomed_code',
+            field=models.IntegerField(blank=True, null=True),
+        ),
+        migrations.AddField(
+            model_name='cohort',
+            name='preexisting_condition_snomed_code',
+            field=models.IntegerField(blank=True, null=True),
+        ),
+        migrations.AddField(
+            model_name='cohort',
+            name='trace',
+            field=models.OneToOneField(null=True, on_delete=django.db.models.deletion.CASCADE, related_name='cohort', to='extraction.trace'),
+        ),
+        migrations.AlterField(
+            model_name='cohort',
+            name='age',
+            field=models.IntegerField(blank=True, null=True),
+        ),
+        migrations.AlterField(
+            model_name='cohort',
+            name='condition',
+            field=models.CharField(blank=True, max_length=50, null=True),
+        ),
+        migrations.AlterField(
+            model_name='cohort',
+            name='origin',
+            field=models.CharField(blank=True, max_length=50, null=True),
+        ),
+        migrations.AlterField(
+            model_name='cohort',
+            name='preexisting_condition',
+            field=models.CharField(blank=True, max_length=100, null=True),
+        ),
+    ]

tracex_project/extraction/migrations/0023_remove_cohort_gender_cohort_sex.py

@@ -0,0 +1,22 @@
+# Generated by Django 4.2.13 on 2024-05-26 09:41
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('extraction', '0022_add_snomed_trace_cohort'),
+    ]
+
+    operations = [
+        migrations.RemoveField(
+            model_name='cohort',
+            name='gender',
+        ),
+        migrations.AddField(
+            model_name='cohort',
+            name='sex',
+            field=models.CharField(blank=True, max_length=25, null=True),
+        ),
+    ]

tracex_project/extraction/models.py

@@ -42,7 +42,9 @@ class Cohort(models.Model):
     sex = models.CharField(max_length=25, null=True, blank=True)
     origin = models.CharField(max_length=50, null=True, blank=True)
     condition = models.CharField(max_length=50, null=True, blank=True)
+    condition_snomed_code = models.IntegerField(null=True, blank=True)
     preexisting_condition = models.CharField(max_length=100, null=True, blank=True)
+    preexisting_condition_snomed_code = models.IntegerField(null=True, blank=True)
     manager = models.Manager()
 
     def __str__(self):

tracex_project/tracex/logic/constants.py

@@ -119,3 +119,16 @@
     ("activity_labeling", "Activity Labeler"),
     ("cohort_tagging", "Cohort Tagger")
 ]
+SNOMED_CT_API_URL = (
+    "https://browser.ihtsdotools.org/snowstorm/snomed-ct/browser/MAIN/descriptions"
+)
+SNOMED_CT_PARAMS = params = {
+    "limit": 1,
+    "conceptActive": "true",
+    "lang": "english",
+    "skipTo": 0,
+    "returnLimit": 1,
+}
+SNOMED_CT_HEADERS = {
+    "User-Agent": "browser",
+}

tracex_project/tracex/logic/utils.py

@@ -10,6 +10,7 @@
 DataFrameUtilities -- Groups all functions related to DataFrame operations.
 """
 import os
+import json
 from pathlib import Path
 import base64
 import tempfile
@@ -19,16 +20,21 @@
 import pandas as pd
 import pm4py
 import numpy as np
+import requests
 
 from django.conf import settings
 from django.db.models import Q
 from openai import OpenAI
+
 from tracex.logic.logger import log_tokens_used
 from tracex.logic.constants import (
     MAX_TOKENS,
     TEMPERATURE_SUMMARIZING,
     MODEL,
     OAIK,
+    SNOMED_CT_API_URL,
+    SNOMED_CT_PARAMS,
+    SNOMED_CT_HEADERS,
 )
 
 from extraction.models import Trace
@@ -120,6 +126,22 @@ def get_snippet_bounds(index: int, length: int) -> tuple[int, int]:
 
     return lower_bound, upper_bound
 
+def get_snomed_ct_info(term):
+    """Get the first matched name and code of a SNOMED CT term."""
+    SNOMED_CT_PARAMS["term"] = term
+    response = requests.get(
+        SNOMED_CT_API_URL, params=SNOMED_CT_PARAMS, headers=SNOMED_CT_HEADERS
+    )
+    data = json.loads(response.text)
+
+    term = None
+    code = None
+
+    if data.get("items"):
+        term = data["items"][0]["term"]
+        code = data["items"][0]["concept"]["conceptId"]
+
+    return term, code
 
 class Conversion:
     """