From b3b4afc53cb9274b5f7bb7318c7185b11eaf6ea2 Mon Sep 17 00:00:00 2001
From: Laura Boenchen <laura.boenchen@limebit.de>
Date: Wed, 11 Sep 2024 13:38:44 +0200
Subject: [PATCH] added interface for comparer and builder

---
 .../comparer/builder.pyi                      | 34 ++++++++++
 .../comparer/data_comparer.pyi                | 68 +++++++++++++++++++
 2 files changed, 102 insertions(+)
 create mode 100644 medmodels/statistic_evaluations/comparer/builder.pyi
 create mode 100644 medmodels/statistic_evaluations/comparer/data_comparer.pyi

diff --git a/medmodels/statistic_evaluations/comparer/builder.pyi b/medmodels/statistic_evaluations/comparer/builder.pyi
new file mode 100644
index 00000000..bce610aa
--- /dev/null
+++ b/medmodels/statistic_evaluations/comparer/builder.pyi
@@ -0,0 +1,34 @@
+from __future__ import annotations
+from typing import List, Optional, Union
+
+from medmodels.medrecord.types import Group, GroupInputList, MedRecordAttribute
+
+from medmodels.statistic_evaluations.comparer.data_comparer import DataComparer
+from medmodels.medrecord.querying import NodeOperation
+
+class ComparerBuilder:
+    patients_group: Optional[Group]
+    time_attribute: Optional[MedRecordAttribute]
+    concept_group: Optional[Union[GroupInputList, Group]]
+    subpopulations_groups: Optional[GroupInputList]
+    subpopulations_queries: Optional[List[NodeOperation]]
+    control_population: Optional[Union[Group, NodeOperation]]
+    top_k_concepts: Optional[int]
+    alpha: Optional[float]
+
+    def with_patients_group(self, group: Group) -> ComparerBuilder: ...
+    def with_time_attributes(
+        self, attribute: MedRecordAttribute
+    ) -> ComparerBuilder: ...
+    def with_concept_group(
+        self, group: Union[Group, GroupInputList]
+    ) -> ComparerBuilder: ...
+    def with_subpopulations(
+        self,
+        groups: GroupInputList,
+        queries: List[NodeOperation],
+        control_population: Union[Group, NodeOperation],
+    ) -> ComparerBuilder: ...
+    def with_top_k(self, k: int) -> ComparerBuilder: ...
+    def with_significance_level(self, alpha: float) -> ComparerBuilder: ...
+    def build(self) -> DataComparer: ...
diff --git a/medmodels/statistic_evaluations/comparer/data_comparer.pyi b/medmodels/statistic_evaluations/comparer/data_comparer.pyi
new file mode 100644
index 00000000..b75dfb8c
--- /dev/null
+++ b/medmodels/statistic_evaluations/comparer/data_comparer.pyi
@@ -0,0 +1,68 @@
+from __future__ import annotations
+
+from typing import Any, Dict, List, Optional, Tuple, Union
+
+import polars as pl
+
+from medmodels.medrecord.medrecord import MedRecord
+from medmodels.medrecord.types import (
+    Group,
+    GroupInputList,
+    MedRecordAttribute,
+    MedRecordAttributeInputList,
+)
+from medmodels.medrecord.querying import NodeOperation
+from medmodels.statistic_evaluations.comparer.builder import ComparerBuilder
+
+class DataComparer:
+
+    patients_group: Group
+    time_attribute: MedRecordAttribute
+    concept_group: Union[GroupInputList, Group]
+    top_k_concepts: int
+    alpha: float
+    subpopulations_groups: Optional[GroupInputList]
+    subpopulations_queries: Optional[List[NodeOperation]]
+    control_population: Optional[Union[Group, NodeOperation]]
+
+    @classmethod
+    def builder(cls) -> ComparerBuilder: ...
+    @staticmethod
+    def _set_configuration(
+        data_comparer: DataComparer,
+        *,
+        patient_group: Group = "patients",
+        time_attribute: MedRecordAttribute = "time",
+        concept_group: Union[GroupInputList, Group] = "concepts",
+        top_k_concepts: int = 10,
+        alpha: float = 0.05,
+        subpopulations_groups: Optional[GroupInputList],
+        subpopulations_queries: Optional[List[NodeOperation]],
+        control_population: Optional[Union[Group, NodeOperation]],
+    ) -> None: ...
+    def compare_attributes(
+        self,
+        medrecords: List[MedRecord],
+        attributes: MedRecordAttributeInputList,
+    ) -> Dict[MedRecordAttribute, pl.DataFrame]: ...
+    def calculate_absolute_relative_difference(
+        self,
+        medrecords: Union[MedRecord, List[MedRecord]],
+        attributes: MedRecordAttributeInputList,
+        control_data: Optional[MedRecord],
+    ) -> Tuple[float, pl.DataFrame]: ...
+    def test_difference_attributes(
+        self,
+        medrecords: Union[MedRecord, List[MedRecord]],
+        attributes: MedRecordAttributeInputList,
+    ) -> Dict[MedRecordAttribute, Tuple[float, float]]: ...
+    def get_top_k_concepts(
+        self,
+        medrecord: MedRecord,
+    ) -> Dict[Union[Group, str], List[Any]]: ...
+    def calculate_distance_concepts(
+        self, medrecord: Union[List[MedRecord], MedRecord]
+    ) -> Dict[str, float]: ...
+    def full_report(
+        self, medrecord: Union[MedRecord, List[MedRecord]]
+    ) -> Dict[str, Any]: ...