vivarium-public-health 3.1.2__tar.gz → 3.1.3__tar.gz

{vivarium_public_health-3.1.2 → vivarium_public_health-3.1.3}/CHANGELOG.rst

@@ -1,3 +1,7 @@
+**3.1.3 - 12/03/24**
+
+ - Feature: Create relative risk pipeline
+
 **3.1.2 - 11/13/24**
 
  - Modernize type hinting

{vivarium_public_health-3.1.2 → vivarium_public_health-3.1.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: vivarium_public_health
-Version: 3.1.2
+Version: 3.1.3
 Summary: Components for modelling diseases, risks, and interventions with ``vivarium``
 Home-page: https://github.com/ihmeuw/vivarium_public_health
 Author: The vivarium developers

vivarium_public_health-3.1.3/src/vivarium_public_health/_version.py

@@ -0,0 +1 @@
+__version__ = "3.1.3"

{vivarium_public_health-3.1.2 → vivarium_public_health-3.1.3}/src/vivarium_public_health/risks/effect.py

@@ -17,6 +17,7 @@ import scipy
 from layered_config_tree import ConfigurationError
 from vivarium import Component
 from vivarium.framework.engine import Builder
+from vivarium.framework.values import Pipeline
 
 from vivarium_public_health.risks import Risk
 from vivarium_public_health.risks.data_transformations import (
@@ -112,7 +113,8 @@ class RiskEffect(Component):
     def setup(self, builder: Builder) -> None:
         self.exposure = self.get_risk_exposure(builder)
 
-        self.target_modifier = self.get_target_modifier(builder)
+        self._relative_risk_source = self.get_relative_risk_source(builder)
+        self.relative_risk = self.get_relative_risk(builder)
 
         self.register_target_modifier(builder)
         self.register_paf_modifier(builder)
@@ -124,7 +126,7 @@ class RiskEffect(Component):
     def build_all_lookup_tables(self, builder: Builder) -> None:
         self._exposure_distribution_type = self.get_distribution_type(builder)
 
-        rr_data = self.get_relative_risk_data(builder)
+        rr_data = self.load_relative_risk(builder)
         rr_value_cols = None
         if self.is_exposure_categorical:
             rr_data, rr_value_cols = self.process_categorical_data(builder, rr_data)
@@ -146,7 +148,7 @@ class RiskEffect(Component):
             return risk_exposure_component.distribution_type
         return risk_exposure_component.get_distribution_type(builder)
 
-    def get_relative_risk_data(
+    def load_relative_risk(
         self,
         builder: Builder,
         configuration=None,
@@ -260,24 +262,27 @@ class RiskEffect(Component):
     def get_risk_exposure(self, builder: Builder) -> Callable[[pd.Index], pd.Series]:
         return builder.value.get_value(self.exposure_pipeline_name)
 
-    def get_target_modifier(
-        self, builder: Builder
-    ) -> Callable[[pd.Index, pd.Series], pd.Series]:
+    def adjust_target(self, index: pd.Index, target: pd.Series) -> pd.Series:
+        relative_risk = self.relative_risk(index)
+        return target * relative_risk
+
+    def get_relative_risk_source(self, builder: Builder) -> Callable[[pd.Index], pd.Series]:
+
         if not self.is_exposure_categorical:
             tmred = builder.data.load(f"{self.risk}.tmred")
             tmrel = 0.5 * (tmred["min"] + tmred["max"])
             scale = builder.data.load(f"{self.risk}.relative_risk_scalar")
 
-            def adjust_target(index: pd.Index, target: pd.Series) -> pd.Series:
+            def generate_relative_risk(index: pd.Index) -> pd.Series:
                 rr = self.lookup_tables["relative_risk"](index)
                 exposure = self.exposure(index)
                 relative_risk = np.maximum(rr.values ** ((exposure - tmrel) / scale), 1)
-                return target * relative_risk
+                return relative_risk
 
         else:
             index_columns = ["index", self.risk.name]
 
-            def adjust_target(index: pd.Index, target: pd.Series) -> pd.Series:
+            def generate_relative_risk(index: pd.Index) -> pd.Series:
                 rr = self.lookup_tables["relative_risk"](index)
                 exposure = self.exposure(index).reset_index()
                 exposure.columns = index_columns
@@ -288,16 +293,24 @@ class RiskEffect(Component):
                 relative_risk = relative_risk.set_index(index_columns)
 
                 effect = relative_risk.loc[exposure.index, "value"].droplevel(self.risk.name)
-                affected_rates = target * effect
-                return affected_rates
+                return effect
 
-        return adjust_target
+        return generate_relative_risk
+
+    def get_relative_risk(self, builder: Builder) -> Pipeline:
+        return builder.value.register_value_producer(
+            f"{self.risk.name}_on_{self.target.name}.relative_risk",
+            self._relative_risk_source,
+            component=self,
+            required_resources=[self.exposure],
+        )
 
     def register_target_modifier(self, builder: Builder) -> None:
         builder.value.register_value_modifier(
             self.target_pipeline_name,
-            modifier=self.target_modifier,
-            requires_values=[f"{self.risk.name}.exposure"],
+            modifier=self.adjust_target,
+            component=self,
+            required_resources=[self.relative_risk],
         )
 
     def register_paf_modifier(self, builder: Builder) -> None:
@@ -307,7 +320,8 @@ class RiskEffect(Component):
         builder.value.register_value_modifier(
             self.target_paf_pipeline_name,
             modifier=self.lookup_tables["population_attributable_fraction"],
-            requires_columns=required_columns,
+            component=self,
+            required_resources=required_columns,
         )
 
     ##################
@@ -371,7 +385,7 @@ class NonLogLinearRiskEffect(RiskEffect):
         return f"non_log_linear_risk_effect.{risk.name}_on_{target}"
 
     def build_all_lookup_tables(self, builder: Builder) -> None:
-        rr_data = self.get_relative_risk_data(builder)
+        rr_data = self.load_relative_risk(builder)
         self.validate_rr_data(rr_data)
 
         def define_rr_intervals(df: pd.DataFrame) -> pd.DataFrame:
@@ -415,7 +429,7 @@ class NonLogLinearRiskEffect(RiskEffect):
             builder, paf_data
         )
 
-    def get_relative_risk_data(
+    def load_relative_risk(
         self,
         builder: Builder,
         configuration=None,
@@ -472,10 +486,8 @@ class NonLogLinearRiskEffect(RiskEffect):
 
         return rr_data
 
-    def get_target_modifier(
-        self, builder: Builder
-    ) -> Callable[[pd.Index, pd.Series], pd.Series]:
-        def adjust_target(index: pd.Index, target: pd.Series) -> pd.Series:
+    def get_relative_risk_source(self, builder: Builder) -> Callable[[pd.Index], pd.Series]:
+        def generate_relative_risk(index: pd.Index) -> pd.Series:
             rr_intervals = self.lookup_tables["relative_risk"](index)
             exposure = self.population_view.get(index)[f"{self.risk.name}_exposure"]
             x1, x2 = (
@@ -486,9 +498,9 @@ class NonLogLinearRiskEffect(RiskEffect):
             m = (y2 - y1) / (x2 - x1)
             b = y1 - m * x1
             relative_risk = b + m * exposure
-            return target * relative_risk
+            return relative_risk
 
-        return adjust_target
+        return generate_relative_risk
 
     ##############
     # Validators #

{vivarium_public_health-3.1.2 → vivarium_public_health-3.1.3}/src/vivarium_public_health/risks/implementations/low_birth_weight_and_short_gestation.py

@@ -336,7 +336,6 @@ class LBWSGRiskEffect(RiskEffect):
 
         super().setup(builder)
         self.interpolator = self.get_interpolator(builder)
-        self.relative_risk = self.get_relative_risk_pipeline(builder)
 
     #################
     # Setup methods #
@@ -393,10 +392,10 @@ class LBWSGRiskEffect(RiskEffect):
             for age_start in exposed_age_group_starts
         }
 
-    def get_relative_risk_pipeline(self, builder: Builder) -> Pipeline:
+    def get_relative_risk(self, builder: Builder) -> Pipeline:
         return builder.value.register_value_producer(
             self.relative_risk_pipeline_name,
-            source=self.get_relative_risk,
+            source=self.get_relative_risk_source,
             requires_columns=["age"] + self.rr_column_names,
         )
 
@@ -470,7 +469,7 @@ class LBWSGRiskEffect(RiskEffect):
     # Pipeline sources and modifiers #
     ##################################
 
-    def get_relative_risk(self, index: pd.Index) -> pd.Series:
+    def get_relative_risk_source(self, index: pd.Index) -> pd.Series:
         pop = self.population_view.get(index)
         relative_risk = pd.Series(1.0, index=index, name=self.relative_risk_pipeline_name)
 

{vivarium_public_health-3.1.2 → vivarium_public_health-3.1.3}/src/vivarium_public_health.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: vivarium_public_health
-Version: 3.1.2
+Version: 3.1.3
 Summary: Components for modelling diseases, risks, and interventions with ``vivarium``
 Home-page: https://github.com/ihmeuw/vivarium_public_health
 Author: The vivarium developers

{vivarium_public_health-3.1.2 → vivarium_public_health-3.1.3}/tests/risks/test_effect.py

@@ -606,3 +606,43 @@ def test_non_loglinear_effect(rr_parameter_data, error_message, base_config, bas
     )
 
     assert np.isclose(rate.values, expected_values, rtol=0.0000001).all()
+
+
+def test_relative_risk_pipeline(dichotomous_risk, base_config, base_plugins):
+    risk = dichotomous_risk[0]
+    effect = RiskEffect(risk.name, "cause.test_cause.incidence_rate")
+    base_config.update({"risk_factor.test_risk": {"data_sources": {"exposure": 0.75}}})
+
+    # TMREL of 1
+    tmred = {"distribution": "uniform", "min": 1, "max": 1, "inverted": False}
+
+    data = {
+        f"{risk.name}.tmred": tmred,
+        f"{risk.name}.population_attributable_fraction": 0,
+        "cause.test_cause.incidence_rate": 1,
+    }
+    rr_value = 1.4
+    base_config.update(
+        {
+            "risk_effect.test_risk_on_cause.test_cause.incidence_rate": {
+                "data_sources": {"relative_risk": rr_value}
+            }
+        }
+    )
+
+    base_config.update({"risk_factor.test_risk": {"distribution_type": "dichotomous"}})
+    sim = _setup_risk_effect_simulation(base_config, base_plugins, risk, effect, data)
+    pop = sim.get_population()
+
+    expected_pipeline_name = f"{effect.risk.name}_on_{effect.target.name}.relative_risk"
+    assert expected_pipeline_name in sim.list_values()
+
+    rr_mapper = {
+        "cat1": 1.4,
+        "cat2": 1.0,
+    }
+    for exposure in rr_mapper:
+        exposure_pipeline = sim.get_value(f"{effect.risk.name}.exposure")(pop.index)
+        exposure_idx = exposure_pipeline.loc[exposure_pipeline == exposure].index
+        relative_risk = sim.get_value(expected_pipeline_name)(exposure_idx)
+        assert (relative_risk == rr_mapper[exposure]).all()

vivarium_public_health-3.1.2/src/vivarium_public_health/_version.py

@@ -1 +0,0 @@
-__version__ = "3.1.2"