Merge branch 'master' of github.com:MIT-LCP/mimic-code

tests/test_postgres_build.py

@@ -5,8 +5,13 @@
 import os
 from subprocess import call
 
+# Prep for Oracle and MySQL database connection
+# http://stackoverflow.com/questions/10065051/python-pandas-and-databases-like-mysql
+# import cx_Oracle
+# import MySQLdb
+
 # Config
-sqluser = 'postgres'
+psqluser = 'postgres'
 testdbname = 'mimic_test_db'
 hostname = 'localhost'
 datadir = 'testdata/v1_3/'
@@ -78,21 +83,41 @@ def run_postgres_build_scripts(cur):
         mimic_data_dir = '/home/mimicadmin/data/mimiciii_1_3/'
     else: 
         mimic_data_dir = curpath+datadir
-    call(['psql','-f',fn,'-d',testdbname,'-U',sqluser,'-v','mimic_data_dir='+mimic_data_dir])
+    call(['psql','-f',fn,'-d',testdbname,'-U',psqluser,'-v','mimic_data_dir='+mimic_data_dir])
     # Add constraints
     fn = curpath + '../buildmimic/postgres/postgres_add_constraints.sql'
     cur.execute(open(fn, "r").read())
     # Add indexes
     fn = curpath + '../buildmimic/postgres/postgres_add_indexes.sql'
     cur.execute(open(fn, "r").read())
 
+# # Prep for adding MySQL build
+# def run_mysql_build_scripts(cur):
+#     # Create tables
+#     fn = curpath + '../buildmimic/mysql/mysql_create_tables.sql'
+#     cur.execute(open(fn, "r").read())
+#     # Loads data
+#     fn = curpath + '../buildmimic/mysql/mysql_load_data.sql'
+#     if os.environ.has_key('USER') and os.environ['USER'] == 'jenkins': 
+#         # use full dataset
+#         mimic_data_dir = '/home/mimicadmin/data/mimiciii_1_3/'
+#     else: 
+#         mimic_data_dir = curpath+datadir
+#     call(['psql','-f',fn,'-d',testdbname,'-U',psqluser,'-v','mimic_data_dir='+mimic_data_dir])
+#     # Add constraints
+#     fn = curpath + '../buildmimic/mysql/mysql_add_constraints.sql'
+#     cur.execute(open(fn, "r").read())
+#     # Add indexes
+#     fn = curpath + '../buildmimic/mysql/mysql_add_indexes.sql'
+#     cur.execute(open(fn, "r").read())
+
 # Class to run unit tests
 class test_postgres(unittest.TestCase):
     # setUpClass runs once for the class
     @classmethod
     def setUpClass(cls):
         # Connect to default postgres database
-        cls.con = psycopg2.connect(dbname='postgres', user=sqluser)
+        cls.con = psycopg2.connect(dbname='postgres', user=psqluser)
         cls.con.set_isolation_level(ISOLATION_LEVEL_AUTOCOMMIT)
         cls.cur = cls.con.cursor()
         # Create test database
@@ -104,7 +129,7 @@ def setUpClass(cls):
         cls.cur.close()
         cls.con.close()
         # Connect to the test database
-        cls.con = psycopg2.connect(dbname=testdbname, user=sqluser)
+        cls.con = psycopg2.connect(dbname=testdbname, user=psqluser)
         cls.con.set_isolation_level(ISOLATION_LEVEL_AUTOCOMMIT)
         cls.cur = cls.con.cursor()
         # Build the test database
@@ -116,7 +141,7 @@ def setUpClass(cls):
     @classmethod
     def tearDownClass(cls):
         # Connect to default postgres database
-        cls.con = psycopg2.connect(dbname='postgres', user=sqluser)
+        cls.con = psycopg2.connect(dbname='postgres', user=psqluser)
         cls.con.set_isolation_level(ISOLATION_LEVEL_AUTOCOMMIT)
         cls.cur = cls.con.cursor()
         # Drop test database
@@ -127,7 +152,7 @@ def tearDownClass(cls):
     # setUp runs once for each test method
     def setUp(self):
         # Connect to the test database
-        self.con = psycopg2.connect(dbname=testdbname, user=sqluser)
+        self.con = psycopg2.connect(dbname=testdbname, user=psqluser)
         self.con.set_isolation_level(ISOLATION_LEVEL_AUTOCOMMIT)
         self.cur = self.con.cursor()
 
@@ -250,6 +275,42 @@ def test_row_counts_are_as_expected(self):
                 queryresult = pd.read_sql_query(query,self.con)
                 self.assertEqual(queryresult.values[0][0],expectedrows)
 
+        def test_age_and_los_is_expected(self):
+            query = \
+            """
+            WITH icuadmissions as (
+                SELECT a.subject_id, a.hadm_id, i.icustay_id, 
+                    a.admittime as hosp_admittime, a.dischtime as hosp_dischtime, 
+                    i.first_careunit, 
+                    DENSE_RANK() over(PARTITION BY a.hadm_id ORDER BY i.intime ASC) as icu_seq,
+                    p.dob, p.dod, i.intime as icu_intime, i.outtime as icu_outtime, 
+                    i.los as icu_los,
+                    round((EXTRACT(EPOCH FROM (a.dischtime-a.admittime))/60/60/24) :: NUMERIC, 4) as hosp_los, 
+                    p.gender, 
+                    round((EXTRACT(EPOCH FROM (a.admittime-p.dob))/60/60/24/365.242) :: NUMERIC, 4) as age_hosp_in,
+                    round((EXTRACT(EPOCH FROM (i.intime-p.dob))/60/60/24/365.242) :: NUMERIC, 4) as age_icu_in,
+                    hospital_expire_flag,
+                    CASE WHEN p.dod IS NOT NULL 
+                        AND p.dod >= i.intime - interval '6 hour'
+                        AND p.dod <= i.outtime + interval '6 hour' THEN 1 
+                        ELSE 0 END AS icu_expire_flag
+                FROM admissions a
+                INNER JOIN icustays i
+                ON a.hadm_id = i.hadm_id
+                INNER JOIN patients p
+                ON a.subject_id = p.subject_id
+                ORDER BY a.subject_id, i.intime)
+            SELECT round(avg(age_icu_in)) as avg_age_icu, 
+                   round(avg(hosp_los)) as avg_los_hosp, 
+                   round(avg(icu_los)) as avg_los_icu
+            FROM icuadmissions;
+            """
+            queryresult = pd.read_sql_query(query,self.con)
+            self.assertEqual(queryresult['avg_age_icu'].values[0][0],65)
+            self.assertEqual(queryresult['avg_los_hosp'].values[0][0],11)
+            self.assertEqual(queryresult['avg_los_icu'].values[0][0],5)
+
+
 def main():
     unittest.main()