fix: unifying name of temporal attributes on example dataset to "time"

limebit · Dec 2, 2024 · 96bacd4 · 96bacd4
1 parent 6de900c
commit 96bacd4
Show file tree

Hide file tree

Showing 7 changed files with 26 additions and 12 deletions.
diff --git a/crates/medmodels-core/src/medrecord/example_dataset/mod.rs b/crates/medmodels-core/src/medrecord/example_dataset/mod.rs
@@ -190,13 +190,13 @@ impl MedRecord {
                         nodes: HashMap::new(),
                         edges: HashMap::from([
                             (
-                                "diagnosis_time".into(),
+                                "time".into(),
                                 (DataType::DateTime, AttributeType::Temporal).into(),
                             ),
                             (
                                 "duration_days".into(),
                                 (
-                                    DataType::Option(Box::new(DataType::Float)),
+                                    DataType::Option(Box::new(DataType::Int)),
                                     AttributeType::Continuous,
                                 )
                                     .into(),
@@ -211,7 +211,7 @@ impl MedRecord {
                         nodes: HashMap::new(),
                         edges: HashMap::from([
                             (
-                                "start_time".into(),
+                                "time".into(),
                                 (DataType::DateTime, AttributeType::Temporal).into(),
                             ),
                             (
@@ -232,7 +232,7 @@ impl MedRecord {
                         nodes: HashMap::new(),
                         edges: HashMap::from([
                             (
-                                "procedure_time".into(),
+                                "time".into(),
                                 (DataType::DateTime, AttributeType::Temporal).into(),
                             ),
                             (

diff --git a/crates/medmodels-core/src/medrecord/example_dataset/synthetic_data/README.md b/crates/medmodels-core/src/medrecord/example_dataset/synthetic_data/README.md
@@ -45,7 +45,7 @@ There are 60 diagnosis instances for the five patients.
 
 - **patient_id:** Linked to the Patient Demographics
 - **diagnosis_code:** Unique code identifiying a specific diagnosis
-- **diagnosis_time:** Date of the diagnosis
+- **time:** Date of the diagnosis
 - **duration_days** Duration in days in which the diagnosis held true
 
 ### 2. `patient_procedure.csv`
@@ -54,7 +54,7 @@ For all five patients there are 50 procedures in the dataset.
 
 - **patient_id:** Linked to the Patient Demographics
 - **procedure_code:** Unique identifier for each procedure
-- **procedure_time:** Exact time and date for the procedure
+- **time:** Exact time and date for the procedure
 - **duration_minutes:** How long the procedure needed in minutes
 
 ### 3. `patient_drug.csv`
@@ -63,6 +63,6 @@ There are also 50 prescribed medications for the five patients combined.
 
 - **patient_id:** Linked to the Patient Demographics
 - **drug_code:** Unique identifier for the medication prescribed
-- **start_time:** Time and date when the medication was prescribed
+- **time:** Time and date when the medication was prescribed
 - **quantity:** how many doses were prescribed
 - **cost:** Cost of the medication order in Dollar
diff --git a/crates/medmodels-core/src/medrecord/example_dataset/synthetic_data/patient_diagnosis.csv b/crates/medmodels-core/src/medrecord/example_dataset/synthetic_data/patient_diagnosis.csv
@@ -1,4 +1,4 @@
-patient_id,diagnosis_code,diagnosis_time,duration_days
+patient_id,diagnosis_code,time,duration_days
 pat_1,diagnosis_82423001,2014-02-06,
 pat_1,diagnosis_73595000,2014-04-08,1113.0
 pat_1,diagnosis_741062008,2014-04-08,371.0

diff --git a/crates/medmodels-core/src/medrecord/example_dataset/synthetic_data/patient_drug.csv b/crates/medmodels-core/src/medrecord/example_dataset/synthetic_data/patient_drug.csv
@@ -1,4 +1,4 @@
-patient_id,drug_code,start_time,quantity,cost
+patient_id,drug_code,time,quantity,cost
 pat_1,drug_856987,2014-04-08T12:54:59Z,3,215.58
 pat_1,drug_856987,2015-04-14T12:54:59Z,3,281.4
 pat_1,drug_856987,2017-04-25T12:54:59Z,3,105.15

diff --git a/crates/medmodels-core/src/medrecord/example_dataset/synthetic_data/patient_procedure.csv b/crates/medmodels-core/src/medrecord/example_dataset/synthetic_data/patient_procedure.csv
@@ -1,4 +1,4 @@
-patient_id,procedure_code,procedure_time,duration_minutes
+patient_id,procedure_code,time,duration_minutes
 pat_1,procedure_710824005,2017-04-25T12:54:59Z,30
 pat_1,procedure_763302001,2018-05-01T15:02:56Z,21
 pat_1,procedure_171207006,2020-05-12T14:05:34Z,11

diff --git a/medmodels/medrecord/tests/test_overview.py b/medmodels/medrecord/tests/test_overview.py
@@ -171,7 +171,7 @@ def query7(edge: EdgeOperand):
         self.assertDictEqual(
             patient_diagnosis,
             {
-                "diagnosis_time": {
+                "time": {
                     "min": datetime(1962, 10, 21, 0, 0),
                     "max": datetime(2024, 4, 12, 0, 0),
                 },

diff --git a/medmodels/medrecord/tests/test_schema.py b/medmodels/medrecord/tests/test_schema.py
@@ -61,7 +61,7 @@ def test_nodes(self):
     def test_edges(self):
         self.assertEqual(
             {
-                "diagnosis_time": (mr.DateTime(), mr.AttributeType.Temporal),
+                "time": (mr.DateTime(), mr.AttributeType.Temporal),
                 "duration_days": (mr.Option(mr.Float()), mr.AttributeType.Continuous),
             },
             self.schema.group("patient_diagnosis").edges,
@@ -236,3 +236,17 @@ def test_eq(self):
         self.assertEqual(mr.AttributeType.Temporal, PyAttributeType.Temporal)
         self.assertNotEqual(mr.AttributeType.Temporal, mr.AttributeType.Categorical)
         self.assertNotEqual(mr.AttributeType.Temporal, PyAttributeType.Categorical)
+
+
+if __name__ == "__main__":
+    run_test = unittest.TestLoader().loadTestsFromTestCase(TestAttributeType)
+    unittest.TextTestRunner(verbosity=2).run(run_test)
+
+    run_test = unittest.TestLoader().loadTestsFromTestCase(TestAttributesSchema)
+    unittest.TextTestRunner(verbosity=2).run(run_test)
+
+    run_test = unittest.TestLoader().loadTestsFromTestCase(TestGroupSchema)
+    unittest.TextTestRunner(verbosity=2).run(run_test)
+
+    run_test = unittest.TestLoader().loadTestsFromTestCase(TestSchema)
+    unittest.TextTestRunner(verbosity=2).run(run_test)