feat: initial commit

.github/workflows/check_smoke_tests.yml

@@ -53,8 +53,8 @@ jobs:
         done
       
         # Run smoke tests and print the response
-        JSON_BODY_1='{"record": {"birth_date": "2053-11-07", "sex": "M", "mrn": "1234567890", "name":[{"family":"Shepard", "given":["John"]}]}}'
-        JSON_BODY_2='{"algorithm": "dibbs-enhanced", "record": {"birth_date": "2000-12-06", "sex": "M", "mrn": "9876543210", "name":[{"family":"Smith", "given":["William"]}]}}'
+        JSON_BODY_1='{"record": {"birth_date": "2053-11-07", "sex": "M", "identifiers":[{"value": "123456789", "type": "MR"}], "name":[{"family":"Shepard", "given":["John"]}]}}'
+        JSON_BODY_2='{"algorithm": "dibbs-enhanced", "record": {"birth_date": "2000-12-06", "sex": "M", "identifiers":[{"value": "9876543210", "type": "MR"}], "name":[{"family":"Smith", "given":["William"]}]}}'
 
         #basic tests
         RESPONSE_1=$(curl -s -X POST http://localhost:8080/link \

docs/site/reference.md

@@ -15,14 +15,6 @@ linkage evaluation phase. The following features are supported:
 
 :   The patient's birthdate (normalized to `YYYY-MM-DD`).
 
-`MRN`
-
-:   The patient's medical record number.
-
-`SSN`
-
-:   The patient's social security number.
-
 `SEX`
 
 :   The patient's sex (normalized to `M`, `F`, or `U` for unknown).
@@ -83,9 +75,13 @@ linkage evaluation phase. The following features are supported:
 
 :   The patient's email address.
 
-`DRIVERS_LICENSE`
+`IDENTIFIER`
+
+:   An identifier for the patient.  Matching on this will check if any identifier type/authority/value combination matches.
 
-:   The patient's driver's license number.
+`IDENTIFIER:<type>`
+
+:   The patient's specific identifier type. For example, `IDENTIFIER:MR` would be the patient's medical record number.  Unlike `IDENTIFIER`, this will ONLY compare values of a specific type.  Valid type codes can be found here http://hl7.org/fhir/R4/v2/0203/index.html.
 
 
 ### Blocking Key Types
@@ -97,10 +93,6 @@ patient data and used during query retrieval. The following blocking key types a
 
 :   The patients birthdate in the format `YYYY-MM-DD`.
 
-`MRN` (ID: **2**)
-
-:   The last 4 characters of a patient's medical record number.
-
 `SEX` (ID: **3**)
 
 :   The patient's sex in the format of `M`, `F`, or `U` for unknown.
@@ -129,6 +121,10 @@ patient data and used during query retrieval. The following blocking key types a
 
 :   The first 4 characters of the patient's email address.
 
+`IDENTIFIER` (ID: **10**)
+
+:  The identifier triplet containing only the type, authority, and last 4 digits of the value
+
 
 ### Evaluation Functions
 

src/recordlinker/assets/initial_algorithms.json

@@ -9,7 +9,7 @@
             {
                 "blocking_keys": [
                     "BIRTHDATE",
-                    "MRN",
+                    "IDENTIFIER",  
                     "SEX"
                 ],
                 "evaluators": [
@@ -75,7 +75,7 @@
             {
                 "blocking_keys": [
                     "BIRTHDATE",
-                    "MRN",
+                    "IDENTIFIER",
                     "SEX"
                 ],
                 "evaluators": [
@@ -106,7 +106,7 @@
                         "CITY": 2.438553006137189,
                         "FIRST_NAME": 6.849475906891162,
                         "LAST_NAME": 6.350720397426025,
-                        "MRN": 0.3051262572525359,
+                        "IDENTIFIER:MR": 0.3051262572525359,
                         "SEX": 0.7510419059643679,
                         "STATE": 0.022376768992488694,
                         "ZIP": 4.975031471124867
@@ -148,7 +148,7 @@
                         "CITY": 2.438553006137189,
                         "FIRST_NAME": 6.849475906891162,
                         "LAST_NAME": 6.350720397426025,
-                        "MRN": 0.3051262572525359,
+                        "IDENTIFIER:MR": 0.3051262572525359,
                         "SEX": 0.7510419059643679,
                         "STATE": 0.022376768992488694,
                         "ZIP": 4.975031471124867

src/recordlinker/hl7/fhir.py

@@ -33,26 +33,18 @@ def fhir_record_to_pii_record(fhir_record: dict) -> schemas.PIIRecord:
         "birthDate": fhir_record.get("birthDate"),
         "sex": fhir_record.get("gender"),
         "address": fhir_record.get("address", []),
-        "mrn": None,
-        "ssn": None,
         "race": None,
         "gender": None,
         "telecom": fhir_record.get("telecom", []),
-        "drivers_license": None,
+        "identifiers": [],
     }
     for identifier in fhir_record.get("identifier", []):
-        for coding in identifier.get("type", {}).get("coding", []):
-            if coding.get("code") == "MR":
-                val["mrn"] = identifier.get("value")
-            elif coding.get("code") == "SS":
-                val["ssn"] = identifier.get("value")
-            elif coding.get("code") == "DL":
-                license_number = identifier.get("value")
-                authority = identifier.get("assigner", {}).get("identifier", {}).get("value", "")  # Assuming `issuer` contains authority info
-                val["drivers_license"] = {
-                    "value": license_number,
-                    "authority": authority
-                }
+        for code in identifier.get("type", {}).get("coding", []):
+            val["identifiers"].append({
+                "value": identifier.get("value"),
+                "type": code.get("code"),
+                "authority": identifier.get("assigner", {}).get("identifier", {}).get("value", ""),
+            })
     for address in val["address"]:
         address["county"] = address.get("district", "")
         for extension in address.get("extension", []):

src/recordlinker/linking/link.py

@@ -51,7 +51,7 @@ def compare(
     details: dict[str, typing.Any] = {"patient.reference_id": str(patient.reference_id)}
     for e in evals:
         # TODO: can we do this check earlier?
-        feature = getattr(schemas.Feature, e.feature, None)
+        feature = schemas.Feature.parse(e.feature)
         if feature is None:
             raise ValueError(f"Invalid comparison field: {e.feature}")
         # Evaluate the comparison function and append the result to the list

src/recordlinker/linking/matchers.py

@@ -179,7 +179,7 @@ def compare_fuzzy_match(
       beyond which to classify the strings as a partial match.
     :return: A float indicating whether the features are a fuzzy match.
     """
-    similarity_measure, threshold = _get_fuzzy_params(str(key), **kwargs)
+    similarity_measure, threshold = _get_fuzzy_params(str(key.attribute), **kwargs)
     comp_func = getattr(rapidfuzz.distance, similarity_measure).normalized_similarity
     for x in record.feature_iter(key):
         for y in patient.record.feature_iter(key):
@@ -203,11 +203,11 @@ def compare_probabilistic_fuzzy_match(
       beyond which to classify the strings as a partial match.
     :return: A float of the score the feature comparison earned.
     """
-    log_odds = kwargs.get("log_odds", {}).get(str(key))
+    log_odds = kwargs.get("log_odds", {}).get(str(key.attribute))
     if log_odds is None:
         raise ValueError(f"Log odds not found for feature {key}")
 
-    similarity_measure, threshold = _get_fuzzy_params(str(key), **kwargs)
+    similarity_measure, threshold = _get_fuzzy_params(str(key.attribute), **kwargs)
     comp_func = getattr(rapidfuzz.distance, similarity_measure).normalized_similarity
     max_score = 0.0
     for x in patient.record.feature_iter(key):

src/recordlinker/models/mpi.py

@@ -121,14 +121,14 @@ class BlockingKey(enum.Enum):
     """
 
     BIRTHDATE = ("BIRTHDATE", 1, "Date of birth as YYYY-MM-DD")
-    MRN = ("MRN", 2, "Last 4 characters of Medical record number")
     SEX = ("SEX", 3, "Sex at birth; M, F or U")
     ZIP = ("ZIP", 4, "5 digital US Postal Code")
     FIRST_NAME = ("FIRST_NAME", 5, "First 4 characters of the first name")
     LAST_NAME = ("LAST_NAME", 6, "First 4 characters of the last name")
     ADDRESS = ("ADDRESS", 7, "First 4 characters of the address")
     PHONE = ("PHONE", 8, "Last 4 characters of the phone number")
     EMAIL = ("EMAIL", 9, "First 4 characters of the email address")
+    IDENTIFIER = ("IDENTIFIER", 10, "Identifier triplet with only last 4 character of the value. Format \"type:authority:value\"")
 
     def __init__(self, value: str, _id: int, description: str):
         self._value = value

src/recordlinker/schemas/__init__.py

@@ -11,6 +11,7 @@
 from .mpi import PatientRef
 from .mpi import PersonRef
 from .pii import Feature
+from .pii import FeatureAttribute
 from .pii import PIIRecord
 from .seed import Cluster
 from .seed import ClusterGroup
@@ -22,6 +23,7 @@
     "AlgorithmPass",
     "AlgorithmSummary",
     "Feature",
+    "FeatureAttribute",
     "PIIRecord",
     "Prediction",
     "LinkInput",

src/recordlinker/schemas/algorithm.py

@@ -13,7 +13,6 @@
 
 from recordlinker.linking import matchers
 from recordlinker.models.mpi import BlockingKey
-from recordlinker.schemas.pii import Feature
 
 
 class Evaluator(pydantic.BaseModel):
@@ -23,10 +22,9 @@ class Evaluator(pydantic.BaseModel):
 
     model_config = pydantic.ConfigDict(from_attributes=True, use_enum_values=True)
 
-    feature: Feature
+    feature: str
     func: matchers.FeatureFunc
 
-
 class AlgorithmPass(pydantic.BaseModel):
     """
     The schema for an algorithm pass record.