do not add the original visit_type and discharge_facility to the cehr…

…bert visit because their orders could be wrong

src/cehrbert/data_generators/hf_data_generator/meds_to_cehrbert_conversion_rules/meds_to_cehrbert_base.py

@@ -69,6 +69,12 @@ def __init__(self, **kwargs):
         self._discharge_matching_rules = self._create_discharge_matching_rules()
         self._text_event_numeric_event_map = {r.code: r for r in self._create_text_event_to_numeric_event_rules()}
 
+    @abstractmethod
+    def _create_visit_matching_rules(self) -> List[str]:
+        raise NotImplementedError(
+            "Must implement the matching rules for identifying the visits other than ED/admission"
+        )
+
     @abstractmethod
     def _create_ed_admission_matching_rules(self) -> List[str]:
         """

src/cehrbert/data_generators/hf_data_generator/meds_to_cehrbert_conversion_rules/meds_to_cehrbert_micmic4.py

@@ -13,6 +13,9 @@ def __init__(self, default_visit_id, **kwargs):
         super().__init__(**kwargs)
         self.default_visit_id = default_visit_id
 
+    def _create_visit_matching_rules(self) -> List[str]:
+        return []
+
     def _create_ed_admission_matching_rules(self) -> List[str]:
         return ["ED_REGISTRATION//", "TRANSFER_TO//ED"]
 

src/cehrbert/data_generators/hf_data_generator/meds_to_cehrbert_conversion_rules/meds_to_cehrbert_omop.py

@@ -8,6 +8,9 @@
 
 class MedsToCehrbertOMOP(MedsToCehrBertConversion):
 
+    def _create_visit_matching_rules(self) -> List[str]:
+        return ["Visit/"]
+
     def _create_ed_admission_matching_rules(self) -> List[str]:
         return ["Visit/ER"]
 

src/cehrbert/data_generators/hf_data_generator/meds_utils.py

@@ -10,7 +10,11 @@
 from datasets import Dataset, DatasetDict, Split
 from transformers.utils import logging
 
-from cehrbert.data_generators.hf_data_generator import DEFAULT_INPATIENT_CONCEPT_ID, UNKNOWN_VALUE
+from cehrbert.data_generators.hf_data_generator import (
+    DEFAULT_ED_CONCEPT_ID,
+    DEFAULT_INPATIENT_CONCEPT_ID,
+    UNKNOWN_VALUE,
+)
 from cehrbert.data_generators.hf_data_generator.hf_dataset import apply_cehrbert_dataset_mapping
 from cehrbert.data_generators.hf_data_generator.hf_dataset_mapping import MedToCehrBertDatasetMapping
 from cehrbert.data_generators.hf_data_generator.meds_to_cehrbert_conversion_rules import MedsToCehrBertConversion
@@ -132,7 +136,7 @@ def convert_one_patient(
         visit_end_datetime = max([b.max_time for b in blocks])
         discharge_facility = (
             next(filter(None, [b.get_discharge_facility() for b in blocks]), None)
-            if visit_type == DEFAULT_INPATIENT_CONCEPT_ID
+            if visit_type in [DEFAULT_INPATIENT_CONCEPT_ID, DEFAULT_ED_CONCEPT_ID]
             else None
         )
         visit_events = list()

src/cehrbert/data_generators/hf_data_generator/patient_block.py

@@ -74,6 +74,7 @@ def __init__(
         self.has_ed_admission = self._has_ed_admission()
         self.has_admission = self._has_admission()
         self.has_discharge = self._has_discharge()
+        self.discharged_to = self.get_discharge_facility()
 
         # Infer the visit_type from the events
         # Admission takes precedence over ED
@@ -82,7 +83,14 @@ def __init__(
         elif self.has_ed_admission:
             self.visit_type = DEFAULT_ED_CONCEPT_ID
         else:
-            self.visit_type = DEFAULT_OUTPATIENT_CONCEPT_ID
+            self.visit_type = self._infer_visit_type()
+
+    def _infer_visit_type(self) -> str:
+        for event in self.events:
+            for matching_rule in self.conversion.get_ed_admission_matching_rules():
+                if re.match(matching_rule, event.code):
+                    return event.code
+        return DEFAULT_OUTPATIENT_CONCEPT_ID
 
     def _has_ed_admission(self) -> bool:
         """
@@ -196,6 +204,9 @@ def get_meds_events(self) -> Iterable[Event]:
         """
         events = []
         for e in self.events:
+            # We only convert the events that are not visit type and discharge facility events
+            if (e.code == self.visit_type) or (self.discharged_to is not None and e.code == self.discharged_to):
+                continue
             events.extend(self._convert_event(e))
         return events