Merge branch 'mimic' of https://github.com/alan-turing-institute/sqlsynthgen into mimic

Author: myyong

config_story.yaml

@@ -85,8 +85,24 @@ src-stats:
           WHERE m.measurement_concept_id IN (21492239, 21492240)
           ) sub
       GROUP BY sub.gender_concept_id;
-      
-
+  - name: avg_measurements_per_visit_hour
+    query: >
+      SELECT
+          AVG(subquery.measurements_per_hour)::float4 AS avg_measurements_per_hour,
+          stddev(subquery.measurements_per_hour)::float4 AS stddev_measurements_per_hour
+      FROM
+      (SELECT
+              COUNT(*)/2
+                / (EXTRACT(EPOCH FROM (v.visit_end_datetime - v.visit_start_datetime)) / 3600.0)
+                AS measurements_per_hour
+      FROM mimic.measurement m
+      JOIN mimic.visit_occurrence v ON m.visit_occurrence_id = v.visit_occurrence_id
+      WHERE m.measurement_concept_id IN (21492239, 21492240)
+      GROUP BY
+              m.person_id,
+              m.visit_occurrence_id,
+              v.visit_start_datetime,
+              v.visit_end_datetime) subquery;
 tables:
   person:
     num_rows_per_pass: 0

person_story.py

@@ -327,8 +327,13 @@ def generate(
     death_row = (yield death) if death else None
     visit_occurrence = yield gen_visit_occurrence(person, death_row, src_stats)
 
+    # abs to avoid negative rates due to random normal variation
+    avg_rate = abs(random_normal(
+        src_stats["avg_measurements_per_visit_hour"][0]['avg_measurements_per_hour'],
+        src_stats["avg_measurements_per_visit_hour"][0]['stddev_measurements_per_hour'] ))
+
     for event in gen_blood_pressure_events(
-        10.0,
+        avg_rate,
         visit_occurrence,
         person,
         src_stats,