omop-constructs 0.2.8__tar.gz → 0.2.10__tar.gz

{omop_constructs-0.2.8 → omop_constructs-0.2.10}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: omop-constructs
-Version: 0.2.8
+Version: 0.2.10
 Summary: For building complex constructs on top of the omop-alchemy library.
 Author: Georgie Kennedy
 Author-email: Georgie Kennedy <georgina.kennedy@unsw.edu.au>

{omop_constructs-0.2.8 → omop_constructs-0.2.10}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "omop-constructs"
-version = "0.2.8"
+version = "0.2.10"
 description = "For building complex constructs on top of the omop-alchemy library."
 readme = "README.md"
 authors = [

omop_constructs-0.2.10/src/omop_constructs/alchemy/events/__init__.py

@@ -0,0 +1,50 @@
+from .dx_linked_event_mv import (
+    WeightDxMV,
+    WeightChangeDxMV,
+    HeightDxMV,
+    BSADxMV,
+    CreatinineClearanceDxMV,
+    EGFRDxMV,
+    FEV1DxMV,
+    DistressThermometerDxMV,
+    ECOGDxMV,
+    SmokingPYHDxMV,
+)
+from .event_factories import measurement_attached_to_condition_episode, episode_relevant_window
+from .event_queries import (
+    weight_query,
+    weight_change_query,
+    height_query,
+    bsa_query,
+    creatinine_clearance_query,
+    est_gfr_query,  
+    fev1_query,
+    distress_thermometer_query,
+    ecog_query,
+    smoking_pyh_query,
+)
+
+__all__ = [
+    "WeightDxMV",
+    "WeightChangeDxMV",         
+    "HeightDxMV",
+    "BSADxMV",
+    "CreatinineClearanceDxMV",
+    "EGFRDxMV",
+    "FEV1DxMV",
+    "DistressThermometerDxMV",
+    "ECOGDxMV",
+    "SmokingPYHDxMV",
+    "measurement_attached_to_condition_episode",
+    "episode_relevant_window",
+    "weight_query",
+    "weight_change_query",
+    "height_query",
+    "bsa_query",
+    "creatinine_clearance_query",
+    "est_gfr_query",
+    "fev1_query",
+    "distress_thermometer_query",
+    "ecog_query",
+    "smoking_pyh_query",
+]

omop_constructs-0.2.10/src/omop_constructs/alchemy/events/dx_linked_event_mv.py

@@ -0,0 +1,168 @@
+import sqlalchemy as sa
+import sqlalchemy.orm as so
+from datetime import date
+from typing import Optional
+from orm_loader.helpers import Base
+from ...core.materialized import MaterializedViewMixin
+from ...core.constructs import register_construct
+
+from .event_queries import (
+    weight_change_dx,
+    weight_dx,
+    height_dx,
+    bsa_dx,
+    creat_dx,
+    egfr_dx,
+    fev1_dx,
+    dtherm_dx,
+    ecog_dx,
+    pyh_dx
+)
+
+class ConditionEpisodeMeasurementCols:
+    __table_args__ = {"extend_existing": True}
+
+    mv_id: so.Mapped[int] = so.mapped_column(primary_key=True)
+
+    person_id: so.Mapped[int] = so.mapped_column(sa.Integer)
+    event_id: so.Mapped[int] = so.mapped_column(sa.Integer)
+    event_date: so.Mapped[date] = so.mapped_column(sa.Date)
+    event_concept_id: so.Mapped[int] = so.mapped_column(sa.Integer)
+    event_label: so.Mapped[Optional[str]] = so.mapped_column(sa.String, nullable=True)
+
+    value_as_number: so.Mapped[Optional[float]] = so.mapped_column(sa.Float, nullable=True)
+    unit_concept_id: so.Mapped[Optional[int]] = so.mapped_column(sa.Integer, nullable=True)
+
+    disease_episode_id: so.Mapped[Optional[int]] = so.mapped_column(sa.Integer, nullable=True)
+    disease_episode_concept_id: so.Mapped[Optional[int]] = so.mapped_column(sa.Integer, nullable=True)
+    disease_episode_label: so.Mapped[Optional[str]] = so.mapped_column(sa.String, nullable=True)
+
+    disease_episode_start_date: so.Mapped[Optional[date]] = so.mapped_column(sa.Date, nullable=True)
+    disease_episode_end_date: so.Mapped[Optional[date]] = so.mapped_column(sa.Date, nullable=True)
+
+    episode_delta_days: so.Mapped[Optional[int]] = so.mapped_column(sa.Integer, nullable=True)
+
+
+@register_construct
+class WeightDxMV(
+    ConditionEpisodeMeasurementCols,
+    MaterializedViewMixin,
+    Base,
+):
+    __mv_name__ = "weight_dx_mv"
+    __mv_select__ = weight_dx.select()
+    __mv_index__ = "person_id"
+    __deps__ = ()
+    __tablename__ = __mv_name__
+
+
+@register_construct
+class WeightChangeDxMV(
+    ConditionEpisodeMeasurementCols,
+    MaterializedViewMixin,
+    Base,
+):
+    __mv_name__ = "weight_change_dx_mv"
+    __mv_select__ = weight_change_dx.select()
+    __mv_index__ = "person_id"
+    __deps__ = ()
+    __tablename__ = __mv_name__
+
+
+@register_construct
+class HeightDxMV(
+    ConditionEpisodeMeasurementCols,
+    MaterializedViewMixin,
+    Base,
+):
+    __mv_name__ = "height_dx_mv"
+    __mv_select__ = height_dx.select()
+    __mv_index__ = "person_id"
+    __deps__ = ()
+    __tablename__ = __mv_name__
+
+
+@register_construct
+class BSADxMV(
+    ConditionEpisodeMeasurementCols,
+    MaterializedViewMixin,
+    Base,
+):
+    __mv_name__ = "bsa_dx_mv"
+    __mv_select__ = bsa_dx.select()
+    __mv_index__ = "person_id"
+    __deps__ = ()
+    __tablename__ = __mv_name__
+
+@register_construct
+class CreatinineClearanceDxMV(
+    ConditionEpisodeMeasurementCols,
+    MaterializedViewMixin,
+    Base,
+):
+    __mv_name__ = "creatinine_clearance_dx_mv"
+    __mv_select__ = creat_dx.select()
+    __mv_index__ = "person_id"
+    __deps__ = ()
+    __tablename__ = __mv_name__
+
+@register_construct
+class EGFRDxMV(
+    ConditionEpisodeMeasurementCols,
+    MaterializedViewMixin,
+    Base,
+):
+    __mv_name__ = "egfr_dx_mv"
+    __mv_select__ = egfr_dx.select()
+    __mv_index__ = "person_id"
+    __deps__ = ()
+    __tablename__ = __mv_name__ 
+
+
+@register_construct
+class FEV1DxMV( 
+    ConditionEpisodeMeasurementCols,
+    MaterializedViewMixin,
+    Base,
+):
+    __mv_name__ = "fev1_dx_mv"
+    __mv_select__ = fev1_dx.select()
+    __mv_index__ = "person_id"
+    __deps__ = ()
+    __tablename__ = __mv_name__
+
+@register_construct
+class DistressThermometerDxMV(
+    ConditionEpisodeMeasurementCols,
+    MaterializedViewMixin,
+    Base,
+):
+    __mv_name__ = "dtherm_dx_mv"
+    __mv_select__ = dtherm_dx.select()
+    __mv_index__ = "person_id"
+    __deps__ = ()
+    __tablename__ = __mv_name__
+
+@register_construct
+class ECOGDxMV(
+    ConditionEpisodeMeasurementCols,
+    MaterializedViewMixin,
+    Base,
+):
+    __mv_name__ = "ecog_dx_mv"
+    __mv_select__ = ecog_dx.select()
+    __mv_index__ = "person_id"
+    __deps__ = ()
+    __tablename__ = __mv_name__
+
+@register_construct
+class SmokingPYHDxMV(
+    ConditionEpisodeMeasurementCols,
+    MaterializedViewMixin,
+    Base,
+):
+    __mv_name__ = "smoking_pyh_dx_mv"
+    __mv_select__ = pyh_dx.select()
+    __mv_index__ = "person_id"
+    __deps__ = ()
+    __tablename__ = __mv_name__

omop_constructs-0.2.10/src/omop_constructs/alchemy/events/event_factories.py

@@ -0,0 +1,195 @@
+import sqlalchemy as sa
+import sqlalchemy.orm as so
+from sqlalchemy.sql import ColumnElement
+from typing import Iterable, Any, Sequence, Union
+from omop_constructs.alchemy.episodes import ConditionEpisodeMV
+from omop_alchemy.cdm.model.structural import Episode_Event
+from omop_semantics.runtime.default_valuesets import runtime # type: ignore
+from omop_alchemy.cdm.model import Measurement, Concept
+
+from sqlalchemy.orm.attributes import InstrumentedAttribute
+SQLExpr = Union[ColumnElement[Any], InstrumentedAttribute[Any]]
+
+modifier_concept = so.aliased(Concept, name="modifier_concept")
+
+def attach_to_condition_episode_via_episode_event(
+    base_event_subq: sa.Subquery,
+    *,
+    event_id_col: ColumnElement[Any],
+    name: str,
+) -> sa.Subquery:
+    """
+    Attach base events to ConditionEpisodeMV via Episode_Event links.
+    """
+    return (
+        sa.select(
+            *base_event_subq.c,
+            ConditionEpisodeMV.disease_episode_id,
+            ConditionEpisodeMV.disease_episode_concept_id,
+            ConditionEpisodeMV.disease_episode_label,
+            ConditionEpisodeMV.disease_episode_start_date,
+            ConditionEpisodeMV.disease_episode_end_date,
+        )
+        .join(
+            Episode_Event,
+            sa.and_(
+                Episode_Event.event_id == event_id_col,
+                Episode_Event.episode_event_field_concept_id
+                == runtime.modifiers.modifier_fields.measurement_id,
+            ),
+        )
+        .join(
+            ConditionEpisodeMV,
+            ConditionEpisodeMV.disease_episode_id == Episode_Event.episode_id,
+        )
+        .subquery(name=name)
+    )   
+
+def attach_to_condition_episode_by_time_window(
+    base_event_subq: sa.Subquery,
+    *,
+    date_col: ColumnElement[Any],
+    person_col: ColumnElement[Any],
+    name: str,
+) -> sa.Subquery:
+    """
+    Time-window attach to ConditionEpisodeMV (fallback when no Episode_Event link exists).
+    """
+    return (
+        sa.select(
+            *base_event_subq.c,
+
+            ConditionEpisodeMV.disease_episode_id,
+            ConditionEpisodeMV.disease_episode_concept_id,
+            ConditionEpisodeMV.disease_episode_label,
+            ConditionEpisodeMV.disease_episode_start_date,
+            ConditionEpisodeMV.disease_episode_end_date,
+            (date_col - ConditionEpisodeMV.disease_episode_start_date).label("episode_delta_days"),
+        )
+        .join(
+            ConditionEpisodeMV,
+            sa.and_(
+                ConditionEpisodeMV.person_id == person_col,
+                date_col >= ConditionEpisodeMV.disease_episode_start_date,
+                sa.or_(
+                    ConditionEpisodeMV.disease_episode_end_date == None,
+                    date_col <= ConditionEpisodeMV.disease_episode_end_date,
+                ),
+            ),
+            isouter=True,
+        )
+        .subquery(name=name)
+    )
+
+def attach_to_condition_episode(
+    base_event_subq: sa.Subquery,
+    *,
+    event_id_col: ColumnElement[Any],
+    date_col: ColumnElement[Any],
+    person_col: ColumnElement[Any],
+    name: str,
+    prefer_explicit_link: bool = True,
+) -> sa.Subquery:
+    """
+    Attach events to ConditionEpisodeMV.
+    
+    Prefers explicit Episode_Event links; can fall back to time-window logic.
+    """
+    explicit = attach_to_condition_episode_via_episode_event(
+        base_event_subq,
+        event_id_col=event_id_col,
+        name=f"{name}_explicit",
+    )
+
+    if not prefer_explicit_link:
+        return explicit
+
+    fallback = attach_to_condition_episode_by_time_window(
+        base_event_subq,
+        date_col=date_col,
+        person_col=person_col,
+        name=f"{name}_fallback",
+    )
+
+    return (
+        sa.select(*explicit.c)
+        .union_all(sa.select(*fallback.c))
+        .subquery(name=name)
+    )
+
+def measurement_event_core(
+    *,
+    concept_ids: Iterable[int] | None = None,
+    name: str = "measurement_core",
+    include_cols: Sequence[SQLExpr] = (),
+    unlinked_only: bool = True,
+) -> sa.Subquery:
+    """
+    Minimal measurement event surface (person/id/date/concept/label + extras).
+    """
+    q = (
+        sa.select(
+            Measurement.person_id.label("person_id"),
+            Measurement.measurement_id.label("event_id"),
+            Measurement.measurement_date.label("event_date"),
+            Measurement.measurement_concept_id.label("event_concept_id"),
+            modifier_concept.concept_name.label("event_label"),
+            *include_cols,
+        )
+        .join(
+            modifier_concept,
+            modifier_concept.concept_id == Measurement.measurement_concept_id,
+            isouter=True,
+        )
+    )
+
+    if unlinked_only:
+        q = q.where(Measurement.modifier_of_event_id == None)
+
+    if concept_ids is not None:
+        q = q.where(Measurement.measurement_concept_id.in_(list(concept_ids)))
+
+    return q.subquery(name=name)
+
+def measurement_attached_to_condition_episode(
+    *,
+    concept_ids: Iterable[int] | None = None,
+    include_cols: Sequence[SQLExpr] = (),
+    name: str,
+    unlinked_only: bool = True,
+) -> sa.Subquery:
+    core = measurement_event_core(
+        concept_ids=concept_ids,
+        include_cols=include_cols,
+        name=f"{name}_core",
+        unlinked_only=unlinked_only,
+    )
+
+    return attach_to_condition_episode(
+        core,
+        event_id_col=core.c.event_id,
+        date_col=core.c.event_date,
+        person_col=core.c.person_id,
+        name=name,
+    )
+
+def episode_relevant_window(
+    starting_query: sa.Subquery,
+    *,
+    max_days_post: int = 90,
+    max_days_prior: int = 30,
+    name: str | None = None,
+) -> sa.Subquery:
+    return (
+        sa.select(
+            sa.func.row_number().over().label("mv_id"),
+            *starting_query.c,
+        )
+        .where(
+            sa.and_(
+                starting_query.c.episode_delta_days <= max_days_post,
+                starting_query.c.episode_delta_days >= -1 * max_days_prior,
+            )
+        )
+        .subquery(name=name or starting_query.name)
+    )

omop_constructs-0.2.10/src/omop_constructs/alchemy/events/event_queries.py

@@ -0,0 +1,75 @@
+from omop_alchemy.cdm.model import Measurement
+from omop_semantics.runtime.default_valuesets import runtime # type: ignore
+from .event_factories import measurement_attached_to_condition_episode, episode_relevant_window
+
+weight_query = measurement_attached_to_condition_episode(
+    concept_ids=[runtime.measurements_numeric.body_size_measurements.weight], # type: ignore
+    include_cols=[Measurement.value_as_number, Measurement.unit_concept_id],
+    name="weight",
+)
+
+weight_change_query = measurement_attached_to_condition_episode(
+    concept_ids=[runtime.measurements_numeric.body_size_measurements.weight_change],  # type: ignore
+    include_cols=[Measurement.value_as_number, Measurement.unit_concept_id],
+    name="weight_change",
+)
+
+height_query = measurement_attached_to_condition_episode(
+    concept_ids=[runtime.measurements_numeric.body_size_measurements.height],  # type: ignore
+    include_cols=[Measurement.value_as_number, Measurement.unit_concept_id],
+    name="height",
+)
+
+bsa_query = measurement_attached_to_condition_episode(
+    concept_ids=[runtime.measurements_numeric.body_size_measurements.bsa],  # type: ignore
+    include_cols=[Measurement.value_as_number, Measurement.unit_concept_id],
+    name="bsa",
+)
+
+creatinine_clearance_query = measurement_attached_to_condition_episode(
+    concept_ids=[runtime.measurements_numeric.lab_measurements.creatinine_clearance],  # type: ignore
+    include_cols=[Measurement.value_as_number, Measurement.unit_concept_id],
+    name="creatinine_clearance",
+)
+
+est_gfr_query = measurement_attached_to_condition_episode(
+    concept_ids=[runtime.measurements_numeric.lab_measurements.egfr],  # type: ignore
+    include_cols=[Measurement.value_as_number, Measurement.unit_concept_id],
+    name="egfr",
+)
+
+fev1_query = measurement_attached_to_condition_episode(
+    concept_ids=[runtime.measurements_numeric.lab_measurements.fev1],  # type: ignore
+    include_cols=[Measurement.value_as_number, Measurement.unit_concept_id],
+    name="fev1",
+)
+
+distress_thermometer_query = measurement_attached_to_condition_episode(
+    concept_ids=[runtime.measurements_numeric.proms_numeric.distress_thermometer],  # type: ignore
+    include_cols=[Measurement.value_as_number],
+    name="dtherm",
+)
+
+ecog_query = measurement_attached_to_condition_episode(
+    concept_ids=[runtime.measurements_numeric.performance_status_measurements.ecog_performance_status],  # type: ignore
+    include_cols=[Measurement.value_as_number],
+    name="ecog",
+)
+
+smoking_pyh_query = measurement_attached_to_condition_episode(
+    concept_ids=[runtime.measurements_numeric.smoking_numeric.pyh],  # type: ignore
+    include_cols=[Measurement.value_as_number],
+    name="smoking_pyh",
+)
+
+
+weight_dx = episode_relevant_window(weight_query, max_days_post=90, max_days_prior=30, name="weight_dx")
+weight_change_dx = episode_relevant_window(weight_change_query, max_days_post=90, max_days_prior=30, name="weight_change_dx")
+height_dx = episode_relevant_window(height_query, max_days_post=90, max_days_prior=30, name="height_dx")
+bsa_dx = episode_relevant_window(bsa_query, max_days_post=90, max_days_prior=30, name="bsa_dx")
+creat_dx = episode_relevant_window(creatinine_clearance_query, max_days_post=90, max_days_prior=30, name="creat_dx")
+egfr_dx = episode_relevant_window(est_gfr_query, max_days_post=90, max_days_prior=30, name="egfr_dx")
+fev1_dx = episode_relevant_window(fev1_query, max_days_post=90, max_days_prior=30, name="fev1_dx")
+dtherm_dx = episode_relevant_window(distress_thermometer_query, max_days_post=90, max_days_prior=30, name="dtherm_dx")
+ecog_dx = episode_relevant_window(ecog_query, max_days_post=90, max_days_prior=30, name="ecog_dx")
+pyh_dx = episode_relevant_window(smoking_pyh_query, max_days_post=90, max_days_prior=30, name="pyh_dx")