Cb/vda 2238 sd 20250831 indexing

underlay/src/main/resources/config/datamapping/sd/entity/bloodPressure/all.sql

@@ -15,7 +15,7 @@ SELECT
     WHEN status_code = '4' THEN 'Normal'
     ELSE 'Unknown'
   END AS status_code_name,
-  CAST(FLOOR(TIMESTAMP_DIFF(xbp.measurement_datetime, p.birth_datetime, DAY) / 365.25) AS INT64) AS age_at_occurrence,
+  CAST(FLOOR(TIMESTAMP_DIFF(xbp.measurement_datetime, p.date_of_birth, DAY) / 365.25) AS INT64) AS age_at_occurrence,
   xbp.visit_occurrence_id,
   vo.visit_concept_id,
   vc.concept_name AS visit_concept_name

underlay/src/main/resources/config/datamapping/sd/entity/bmi/all.sql

@@ -16,7 +16,7 @@ SELECT
   mvc.concept_name AS value_as_concept_name,
   mo.unit_concept_id,
   muc.concept_name AS unit_concept_name,
-  CAST(FLOOR(TIMESTAMP_DIFF(mo.measurement_datetime, p.birth_datetime, DAY) / 365.25) AS INT64) AS age_at_occurrence,
+  CAST(FLOOR(TIMESTAMP_DIFF(mo.measurement_datetime, p.date_of_birth, DAY) / 365.25) AS INT64) AS age_at_occurrence,
   mo.visit_occurrence_id,
   vo.visit_concept_id,
   vc.concept_name AS visit_concept_name

underlay/src/main/resources/config/datamapping/sd/entity/conditionOccurrence/all.sql

@@ -11,7 +11,7 @@ SELECT DISTINCT
   (CASE WHEN cc.standard_concept IS NULL THEN 'Source' WHEN cc.standard_concept = 'S' THEN 'Standard' ELSE 'Unknown' END) AS standard_concept,
   co.condition_source_value,
   co.condition_source_concept_id,
-  CAST(FLOOR(TIMESTAMP_DIFF(co.condition_start_datetime, p.birth_datetime, DAY) / 365.25) AS INT64) AS age_at_occurrence,
+  CAST(FLOOR(TIMESTAMP_DIFF(co.condition_start_datetime, p.date_of_birth, DAY) / 365.25) AS INT64) AS age_at_occurrence,
   co.visit_occurrence_id,
   vo.visit_concept_id,
   vc.concept_name AS visit_concept_name

underlay/src/main/resources/config/datamapping/sd/entity/deviceOccurrence/all.sql

@@ -8,7 +8,7 @@ SELECT
   de.device_exposure_end_date,
   de.device_source_value,
   de.device_source_concept_id,
-  CAST(FLOOR(TIMESTAMP_DIFF(de.device_exposure_start_datetime, p.birth_datetime, DAY) / 365.25) AS INT64) AS age_at_occurrence,
+  CAST(FLOOR(TIMESTAMP_DIFF(de.device_exposure_start_datetime, p.date_of_birth, DAY) / 365.25) AS INT64) AS age_at_occurrence,
   de.visit_occurrence_id,
   vo.visit_concept_id,
   vc.concept_name AS visit_concept_name

underlay/src/main/resources/config/datamapping/sd/entity/familyHistory/all.sql

@@ -6,7 +6,7 @@ SELECT
   xfh.item AS note_text, 
   xfh.family_history_src_name, 
   xfh.relation_name,
-  CAST(FLOOR(TIMESTAMP_DIFF(xfh.entry_date, p.birth_datetime, DAY) / 365.25) AS INT64) AS age_at_occurrence,
+  CAST(FLOOR(TIMESTAMP_DIFF(xfh.entry_date, p.date_of_birth, DAY) / 365.25) AS INT64) AS age_at_occurrence,
   xfh.visit_occurrence_id, 
   vo.visit_concept_id,
   vc.concept_name AS visit_concept_name

underlay/src/main/resources/config/datamapping/sd/entity/height/all.sql

@@ -16,7 +16,7 @@ SELECT
   mvc.concept_name AS value_as_concept_name,
   mo.unit_concept_id,
   muc.concept_name AS unit_concept_name,
-  CAST(FLOOR(TIMESTAMP_DIFF(mo.measurement_datetime, p.birth_datetime, DAY) / 365.25) AS INT64) AS age_at_occurrence,
+  CAST(FLOOR(TIMESTAMP_DIFF(mo.measurement_datetime, p.date_of_birth, DAY) / 365.25) AS INT64) AS age_at_occurrence,
   mo.visit_occurrence_id,
   vo.visit_concept_id,
   vc.concept_name AS visit_concept_name

underlay/src/main/resources/config/datamapping/sd/entity/ingredientOccurrence/all.sql

@@ -11,7 +11,7 @@ SELECT
   de.days_supply,
   de.drug_source_value,
   de.drug_source_concept_id,
-  CAST(FLOOR(TIMESTAMP_DIFF(de.drug_exposure_start_datetime, p.birth_datetime, DAY) / 365.25) AS INT64) AS age_at_occurrence,
+  CAST(FLOOR(TIMESTAMP_DIFF(de.drug_exposure_start_datetime, p.date_of_birth, DAY) / 365.25) AS INT64) AS age_at_occurrence,
   de.visit_occurrence_id,
   de.drug_type_concept_id,
   tc.concept_name AS drug_type_concept_name,

underlay/src/main/resources/config/datamapping/sd/entity/measurementOccurrence/all.sql

@@ -12,7 +12,7 @@ SELECT
   uc.concept_name AS unit_concept_name,
   mo.measurement_source_value,
   mo.measurement_source_concept_id,
-  CAST(FLOOR(TIMESTAMP_DIFF(mo.measurement_datetime, p.birth_datetime, DAY) / 365.25) AS INT64) AS age_at_occurrence,
+  CAST(FLOOR(TIMESTAMP_DIFF(mo.measurement_datetime, p.date_of_birth, DAY) / 365.25) AS INT64) AS age_at_occurrence,
   mo.visit_occurrence_id,
   vo.visit_concept_id,
   vc.concept_name AS visit_concept_name,

underlay/src/main/resources/config/datamapping/sd/entity/measurementVumcVocab/all.sql

@@ -5,9 +5,9 @@ SELECT
     c.concept_code,
     -- standard_concept is empty in concept table
     'Source' AS standard_concept
-FROM (SELECT DISTINCT measurement_concept_id FROM `${omopDataset}.measurement`
-      WHERE measurement_concept_id IS NOT NULL
-        AND measurement_concept_id != 0) m
-JOIN `${omopDataset}.concept` c ON m.measurement_concept_id = c.concept_id
+FROM (SELECT DISTINCT measurement_source_concept_id FROM `${omopDataset}.measurement`
+      WHERE measurement_source_concept_id IS NOT NULL
+        AND measurement_concept_id = 0) m
+JOIN `${omopDataset}.concept` c ON m.measurement_source_concept_id = c.concept_id
 WHERE c.domain_id = 'Measurement'
   AND c.vocabulary_id = 'VUMC Meas Value'

underlay/src/main/resources/config/datamapping/sd/entity/noteOccurrence/all.sql

@@ -8,7 +8,7 @@ SELECT
   n.note_title,
   n.note_text,
   n.note_source_value,
-  CAST(FLOOR(TIMESTAMP_DIFF(n.note_datetime, p.birth_datetime, DAY) / 365.25) AS INT64) AS age_at_occurrence,
+  CAST(FLOOR(TIMESTAMP_DIFF(n.note_datetime, p.date_of_birth, DAY) / 365.25) AS INT64) AS age_at_occurrence,
   n.visit_occurrence_id,
   vo.visit_concept_id,
   vc.concept_name AS visit_concept_name

underlay/src/main/resources/config/datamapping/sd/entity/observationOccurrence/all.sql

@@ -14,7 +14,7 @@ SELECT DISTINCT
   (CASE WHEN oc.standard_concept IS NULL THEN 'Source' WHEN oc.standard_concept = 'S' THEN 'Standard' ELSE 'Unknown' END) AS standard_concept,
   o.observation_source_value,
   o.observation_source_concept_id,
-  CAST(FLOOR(TIMESTAMP_DIFF(o.observation_date, p.birth_datetime, DAY) / 365.25) AS INT64) AS age_at_occurrence,
+  CAST(FLOOR(TIMESTAMP_DIFF(o.observation_date, p.date_of_birth, DAY) / 365.25) AS INT64) AS age_at_occurrence,
   o.visit_occurrence_id,
   vo.visit_concept_id,
   vc.concept_name AS visit_concept_name

underlay/src/main/resources/config/datamapping/sd/entity/procedureOccurrence/all.sql

@@ -9,7 +9,7 @@ SELECT
   po.procedure_date,
   po.procedure_source_value,
   po.procedure_source_concept_id,
-  CAST(FLOOR(TIMESTAMP_DIFF(po.procedure_date, p.birth_datetime, DAY) / 365.25) AS INT64) AS age_at_occurrence,
+  CAST(FLOOR(TIMESTAMP_DIFF(po.procedure_date, p.date_of_birth, DAY) / 365.25) AS INT64) AS age_at_occurrence,
   po.visit_occurrence_id,
   vo.visit_concept_id,
   vc.concept_name AS visit_concept_name

underlay/src/main/resources/config/datamapping/sd/entity/pulse/all.sql

@@ -8,7 +8,7 @@ SELECT
   mvc.concept_name AS value_as_concept_name,
   mo.unit_concept_id,
   muc.concept_name AS unit_concept_name,
-  CAST(FLOOR(TIMESTAMP_DIFF(mo.measurement_datetime, p.birth_datetime, DAY) / 365.25) AS INT64) AS age_at_occurrence,
+  CAST(FLOOR(TIMESTAMP_DIFF(mo.measurement_datetime, p.date_of_birth, DAY) / 365.25) AS INT64) AS age_at_occurrence,
   mo.visit_occurrence_id,
   vo.visit_concept_id,
   vc.concept_name AS visit_concept_name

underlay/src/main/resources/config/datamapping/sd/entity/respiratoryRate/all.sql

@@ -8,7 +8,7 @@ SELECT
   mvc.concept_name AS value_as_concept_name,
   mo.unit_concept_id,
   muc.concept_name AS unit_concept_name,
-  CAST(FLOOR(TIMESTAMP_DIFF(mo.measurement_datetime, p.birth_datetime, DAY) / 365.25) AS INT64) AS age_at_occurrence,
+  CAST(FLOOR(TIMESTAMP_DIFF(mo.measurement_datetime, p.date_of_birth, DAY) / 365.25) AS INT64) AS age_at_occurrence,
   mo.visit_occurrence_id,
   vo.visit_concept_id,
   vc.concept_name AS visit_concept_name

underlay/src/main/resources/config/datamapping/sd/entity/visitOccurrence/all.sql

@@ -8,7 +8,7 @@ SELECT
   vo.visit_end_date,
   vo.visit_source_value,
   vo.visit_source_concept_id,
-  CAST(FLOOR(TIMESTAMP_DIFF(vo.visit_start_datetime, p.birth_datetime, DAY) / 365.25) AS INT64) AS age_at_occurrence
+  CAST(FLOOR(TIMESTAMP_DIFF(vo.visit_start_datetime, p.date_of_birth, DAY) / 365.25) AS INT64) AS age_at_occurrence
 FROM `${omopDataset}.visit_occurrence` AS vo
 JOIN `${omopDataset}.person_ext` AS p
     ON p.person_id = vo.person_id

underlay/src/main/resources/config/datamapping/sd/entity/weight/all.sql

@@ -16,7 +16,7 @@ SELECT
   mvc.concept_name AS value_as_concept_name,
   mo.unit_concept_id,
   muc.concept_name AS unit_concept_name,
-  CAST(FLOOR(TIMESTAMP_DIFF(mo.measurement_datetime, p.birth_datetime, DAY) / 365.25) AS INT64) AS age_at_occurrence,
+  CAST(FLOOR(TIMESTAMP_DIFF(mo.measurement_datetime, p.date_of_birth, DAY) / 365.25) AS INT64) AS age_at_occurrence,
   mo.visit_occurrence_id,
   vo.visit_concept_id,
   vc.concept_name AS visit_concept_name

underlay/src/main/resources/config/indexer/sd/sd_20250831.json

@@ -0,0 +1,26 @@
+{
+  "underlay": "sd",
+  "bigQuery": {
+    "sourceData": {
+      "projectId": "vumc-sd-dev",
+      "datasetId": "sd_20250831",
+      "sqlSubstitutions": {
+        "omopDataset": "vumc-sd-dev.sd_20250831",
+        "staticTablesDataset": "vumc-sd-dev.sd_20250831"
+      }
+    },
+    "indexData": {
+      "projectId": "vumc-sd-dev",
+      "datasetId": "indexed_sd_20250831"
+    },
+    "queryProjectId": "vumc-sd-dev",
+    "dataLocation": "us-central1"
+  },
+  "dataflow": {
+    "serviceAccountEmail": "[email protected]",
+    "gcsTempDirectory": "gs://dataflow-indexing-sd-dev/temp/",
+    "workerMachineType": "n1-standard-4",
+    "usePublicIps": false,
+    "vpcSubnetworkName": "us-central1"
+  }
+}

underlay/src/main/resources/config/service/sd/sd_20250831_dev.json

@@ -0,0 +1,21 @@
+{
+  "underlay": "sd",
+  "bigQuery": {
+    "sourceData": {
+      "projectId": "vumc-sd-dev",
+      "datasetId": "sd_20250831",
+      "sqlSubstitutions": {
+        "indexedDataset": "vumc-sd-dev.indexed_sd_20250831",
+        "omopDataset": "vumc-sd-dev.sd_20250831",
+        "staticTablesDataset": "vumc-sd-dev.sd_20250831"
+      }
+    },
+    "indexData": {
+      "projectId": "vumc-sd-dev",
+      "datasetId": "indexed_sd_20250831"
+    },
+    "queryProjectId": "vumc-sd-dev",
+    "dataLocation": "us-central1"
+  },
+  "uiConfigFile": "ui.json"
+}

underlay/src/main/resources/config/service/sd/sd_20250831_prod.json

@@ -0,0 +1,21 @@
+{
+  "underlay": "sd",
+  "bigQuery": {
+    "sourceData": {
+      "projectId": "vumc-sd-prod",
+      "datasetId": "sd_20250228",
+      "sqlSubstitutions": {
+        "indexedDataset": "vumc-sd-prod.indexed_sd_20250831",
+        "omopDataset": "vumc-sd-prod.sd_20250831",
+        "staticTablesDataset": "vumc-sd-prod.sd_20250831"
+      }
+    },
+    "indexData": {
+      "projectId": "vumc-sd-prod",
+      "datasetId": "indexed_sd_20250831"
+    },
+    "queryProjectId": "vumc-sd-prod",
+    "dataLocation": "us-central1"
+  },
+  "uiConfigFile": "ui.json"
+}

underlay/src/main/resources/config/service/sd/sd_20250831_test.json

@@ -0,0 +1,21 @@
+{
+  "underlay": "sd",
+  "bigQuery": {
+    "sourceData": {
+      "projectId": "vumc-sd-test",
+      "datasetId": "sd_20250831",
+      "sqlSubstitutions": {
+        "indexedDataset": "vumc-sd-test.indexed_sd_20250831",
+        "omopDataset": "vumc-sd-test.sd_20250831",
+        "staticTablesDataset": "vumc-sd-test.sd_20250831"
+      }
+    },
+    "indexData": {
+      "projectId": "vumc-sd-test",
+      "datasetId": "indexed_sd_20250831"
+    },
+    "queryProjectId": "vumc-sd-test",
+    "dataLocation": "us-central1"
+  },
+  "uiConfigFile": "ui.json"
+}