Simplify LBWSG category parsing (#546)

* use re findall

* add error

* change check

* isort

* CL update

Author: patricktnast

CHANGELOG.rst

@@ -1,3 +1,7 @@
+**4.3.11 - 09/24/25**
+
+  - Refactor: Simplify LBWSG parsing logic
+
 **4.3.10 - 09/23/25**
 
   - Bugfix: Fix bug in PublicHealthObserver to retain stratification columns in results

src/vivarium_public_health/risks/implementations/low_birth_weight_and_short_gestation.py

@@ -9,6 +9,7 @@
 """
 
 import pickle
+import re
 from collections.abc import Callable
 from typing import Any
 
@@ -119,13 +120,12 @@ def get_category_intervals(self, builder: Builder) -> dict[str, dict[str, pd.Int
             The intervals for each category.
         """
         categories: dict[str, str] = builder.data.load(f"{self.risk}.categories")
-        category_intervals = {
-            axis: {
-                category: self._parse_description(axis, description)
-                for category, description in categories.items()
-            }
-            for axis in [BIRTH_WEIGHT, GESTATIONAL_AGE]
-        }
+        category_intervals = {GESTATIONAL_AGE: {}, BIRTH_WEIGHT: {}}
+
+        for category, description in categories.items():
+            gestation_interval, birth_weight_interval = self._parse_description(description)
+            category_intervals[GESTATIONAL_AGE][category] = gestation_interval
+            category_intervals[BIRTH_WEIGHT][category] = birth_weight_interval
         return category_intervals
 
     ##################
@@ -224,7 +224,7 @@ def single_axis_ppf(
     ##################
 
     @staticmethod
-    def _parse_description(axis: str, description: str) -> pd.Interval:
+    def _parse_description(description: str) -> tuple[pd.Interval, pd.Interval]:
         """Parses a string corresponding to a low birth weight and short gestation
         category to an Interval.
 
@@ -235,17 +235,15 @@ def _parse_description(axis: str, description: str) -> pd.Interval:
         An example of an edge case of birth weight:
         'Neonatal preterm and LBWSG (estimation years) - [36, 37) wks, [4000, 9999] g'
         """
-        endpoints = {
-            BIRTH_WEIGHT: [
-                float(val)
-                for val in description.split(", [")[1].split(")")[0].split("]")[0].split(", ")
-            ],
-            GESTATIONAL_AGE: [
-                float(val)
-                for val in description.split("- [")[1].split(")")[0].split("+")[0].split(", ")
-            ],
-        }[axis]
-        return pd.Interval(*endpoints, closed="left")  # noqa
+        lbwsg_values = [float(val) for val in re.findall(r"(\d+)", description)]
+        if len(list(lbwsg_values)) != 4:
+            raise ValueError(
+                f"Could not parse LBWSG description '{description}'. Expected 4 numeric values."
+            )
+        return (
+            pd.Interval(*lbwsg_values[:2], closed="left"),  # Gestational Age
+            pd.Interval(*lbwsg_values[2:], closed="left"),  # Birth Weight
+        )
 
 
 class LBWSGRisk(Risk):

tests/risks/test_low_birth_weight_and_short_gestation.py

@@ -36,8 +36,7 @@
     ],
 )
 def test_parsing_lbwsg_descriptions(description, expected_weight_values, expected_age_values):
-    weight_interval = LBWSGDistribution._parse_description("birth_weight", description)
-    age_interval = LBWSGDistribution._parse_description("gestational_age", description)
+    age_interval, weight_interval = LBWSGDistribution._parse_description(description)
     assert weight_interval.left == expected_weight_values[0]
     assert weight_interval.right == expected_weight_values[1]
     assert age_interval.left == expected_age_values[0]