add mimic_core view

Author: antheas

src/pasteur/extras/__init__.py

@@ -16,7 +16,7 @@ def get_recommended_datasets() -> list[Dataset | View]:
     from .datasets.mimic import MimicDataset
     from .datasets.texas import TexasDataset
     from .views.adult import TabAdultView
-    from .views.mimic import MimicMmCoreView, MimicTabAdmissions
+    from .views.mimic import MimicCore, MimicTabAdmissions
     # from .views.texas import TexasChargesView, TexasBaseView, TexasBillionView
 
     return [
@@ -27,7 +27,7 @@ def get_recommended_datasets() -> list[Dataset | View]:
         # TexasChargesView(),
         # TexasBaseView(),
         TabAdultView(),
-        # MimicMmCoreView,
+        # MimicCore,
         MimicTabAdmissions(),
         # TexasBillionView(),
     ]

src/pasteur/extras/views/mimic/__init__.py

@@ -30,10 +30,10 @@ def mm_core_transform_patients(patients: pd.DataFrame):
     return patients_new
 
 
-class MimicMmCoreView(View):
+class MimicCore(View):
     """The mimic core tables, slightly post processed."""
 
-    name = "mimic_mm_core"
+    name = "mimic_core"
     dataset = "mimic"
     deps: dict[str, list[str]] = {
         "patients": ["core_patients"],
@@ -44,7 +44,7 @@ class MimicMmCoreView(View):
         "admissions": ["patients"],
         "transfers": ["admissions"],
     }
-    parameters = get_relative_fn("parameters_mm.yml")
+    parameters = get_relative_fn("parameters_core.yml")
 
     @to_chunked
     def ingest(self, name, **tables: LazyChunk):

…eur/extras/views/mimic/parameters_mm.yml …r/extras/views/mimic/parameters_core.ymlsrc/pasteur/extras/views/mimic/parameters_mm.yml renamed to src/pasteur/extras/views/mimic/parameters_core.yml src/pasteur/extras/views/mimic/parameters_mm.yml renamed to src/pasteur/extras/views/mimic/parameters_core.yml

@@ -5,22 +5,28 @@ tables:
       subject_id: id
       gender: categorical
       # year_group: categorical
-      dod: date?:birth_year
+      dod:
+        type: date
+        span: year
+        ref: birth_year
+        nullable: True
+        max_len: 99
+        bins: 32
       birth_year: fixed|date
 
   admissions:
     primary_key: hadm_id
+    sequencer: admittime
+
     fields:
       hadm_id: id
       subject_id: id:patients.subject_id
       admittime:
-        type: datetime
+        type: seqdatetime
         span: year.hour
-        ref:
-          table: patients
-          col: birth_year
+        ref: patients.birth_year
         max_len: 99
-        bins: 100
+        bins: 32
       dischtime: datetime|day.hour:admittime
       deathtime: datetime?|day.hour:admittime
       admission_type: categorical
@@ -36,10 +42,11 @@ tables:
 
   transfers:
     primary_key: transfer_id
+    sequencer: intime
     fields:
       transfer_id: id
       hadm_id: id:admissions.hadm_id
       eventtype: categorical
       careunit: categorical?
-      intime: datetime|day.hour:admissions.admittime
+      intime: seqdatetime|day.hour:admissions.admittime
       outtime: datetime?|day.hour:intime

src/pasteur/table.py

@@ -35,7 +35,6 @@
 
 A = TypeVar("A", bound="Any")
 META = TypeVar("META")
-_IDKEY = "__ids_lkjhasndsfnewr"
 
 
 def _reduce_inner(
@@ -231,13 +230,9 @@ def fit(
         tables: dict[str, LazyFrame],
         ids: LazyFrame | None = None,
     ):
-        if ids is not None:
-            tables = {_IDKEY: ids, **tables}
-
         per_call = []
-        for _, chunks in LazyFrame.zip(tables).items():
-            ids_chunk = chunks.pop(_IDKEY, None)
-            per_call.append({"ids": ids_chunk, "tables": chunks})
+        for cids, ctables in LazyFrame.zip_values([ids, tables]):
+            per_call.append({"ids": cids, "tables": ctables})
 
         transformer_chunks: list[
             dict[str | tuple[str], Transformer]

src/project/settings.py

@@ -53,10 +53,11 @@
 
 from pasteur.extras import get_recommended_modules
 from pasteur.extras.synth.pgm import AIM, MST
-from pasteur.extras.views.mimic import MimicBillion
+from pasteur.extras.views.mimic import MimicBillion, MimicCore
 
 PASTEUR_MODULES = get_recommended_modules() + [
     AIM.get_factory(),
     MST.get_factory(),
     MimicBillion(),
+    MimicCore(),
 ]