Add in vitals conversion.

scripts/transform_vitals.py

+#!/usr/bin/env python3
+
+'''Generate Vitals'''
+
+import argparse
+import os
+import re
+import pandas as pd
+
+
+
+EPILOG = '''
+For more details:
+        %(prog)s --help
+'''
+
+
+def get_args():
+    '''Define arguments.'''
+
+    parser = argparse.ArgumentParser(
+        description=__doc__, epilog=EPILOG,
+        formatter_class=argparse.RawDescriptionHelpFormatter)
+
+    parser.add_argument('-f', '--file',
+                        help="Vitals File (csv format).",
+                        required=True)
+
+    parser.add_argument('-u', '--unitmap',
+                        help="Labs unitmap (csv format).",
+                        required=True)
+
+    parser.add_argument('-d', '--date',
+                        help="Date Shift (tsv format).",
+                        required=True)
+
+    parser.add_argument('-o', '--out',
+                        help="The output path (csv format).",
+                        required=True)
+
+    args = parser.parse_args()
+    return args
+
+
+def filter_recorded(vitals, unitmap):
+    '''Filter for recorded vitals values'''
+
+    # Convert CONCEPT_CD to remove VISIT
+    vitals['CONCEPT_CD'] = pd.DataFrame(list(vitals['CONCEPT_CD'].str.split(':')))[1]
+
+    # Remove missing and dummy values
+    df_clean = vitals[~(vitals.NVAL_NUM.isna() | (vitals.NVAL_NUM == 9999999))]
+
+    # Filter for only vitals we are tracking
+    df_cleaned_tracked = df_clean[(df_clean.CONCEPT_CD).isin(unitmap.Vitals)]
+
+    vital_columns = ["mrn", "vital", "date",
+                   "value", "value_units"]
+    df_cleaned_tracked.columns = vital_columns
+
+    return df_cleaned_tracked
+
+
+def remove_outliers(vitals, measurement):
+    '''Remove outliers for a specific measurement'''
+    ##TODO: Make this patient specific
+
+    # Calculate extremes
+    specific_vitals = vitals[vitals.vital == measurement]
+    q1 = specific_vitals['value'].quantile(0.25)
+    q3 = specific_vitals['value'].quantile(0.75)
+    iqr = q3 - q1
+    upper = (iqr * 3) + q3
+    lower = q1 - (iqr * 3)
+
+    # Filter for extremes
+    filter_df = vitals[-((vitals.vital == measurement) & ((vitals.value > upper) | (vitals.value < lower)))]
+
+    return filter_df
+
+
+def calculate_shift(vitals, date_shift):
+    '''Shift Date for start date'''
+
+    # Convert to DateTime
+    date_shift['Shift'] = pd.to_timedelta(date_shift['Shift'], unit='s')
+    vitals['date'] = pd.to_datetime(vitals['date'])
+
+    # Merge data
+    merged = vitals.merge(date_shift, left_on='mrn', right_on='MRN', how='inner')
+
+    # Calculate Date Shift
+    merged['date'] = merged['date'] + merged['Shift']
+    merged['date'] = merged['date'].dt.date
+
+    # Drop Shift column
+    merged.drop(['Shift'], axis=1, inplace=True)
+    merged.drop(['MRN'], axis=1, inplace=True)
+
+    # Int MRN
+    merged.mrn = merged.mrn.astype(int)
+
+    return merged
+
+
+def main():
+    args = get_args()
+    vitals = args.file
+    unitmap = args.unitmap
+    date = args.date
+    out_path = args.out
+
+    # Make output files
+    vital_table = os.path.join(out_path + 'vital_table.csv')
+
+    # Read in files
+    vitals_df = pd.read_csv(vitals)
+    unitmap_df = pd.read_csv(unitmap)
+    date_shift = pd.read_csv(date)
+
+    # Filter for true records data
+    vitals_filtered = filter_recorded(vitals_df, unitmap_df)
+
+
+    # Remove outliers
+    ##TODO: Will need to update this code
+    #for m in unitmap_df['Vitals']:
+    #  temp = remove_outliers(vitals_filtered, m)
+    #  vitals_filtered = temp
+
+    # Calculate Date Shift
+    shifted_df = calculate_shift(vitals_filtered, date_shift)
+
+    # Write out lab table
+    shifted_df.to_csv(vital_table, index=False)
+
+
+if __name__ == '__main__':
+    main()

standards/vital_unitmap.csv

+Vitals,Units
+BMI,kg/m2
+BP_DIAS,mm Hg
+BP_SYS,mm Hg
+HEIGHT,Inches
+WEIGHT,Pounds

tests/test_transform_vitals.py

+#!/usr/bin/env python3
+
+import pytest
+import os
+import pandas as pd
+from StringIO import StringIO
+import transform_vitals
+
+
+VITALS_STRING = """MRN,CONCEPT_CD,START_DATE,NVAL_NUM,UNITS_CD
+33,VISIT:WEIGHT,9/14/2013 0:00,172,Pounds"
+"""
+
+standards_path = os.path.dirname(os.path.abspath(__file__)) + \
+                '/../standards/'
+
+
+@pytest.fixture
+def vital():
+    vitals_file = StringIO(VITALS_STRING)
+    vitals_df = pd.read_csv(vitals_file)
+    return vitals_df
+
+
+@pytest.fixture
+def unit_map():
+    unit_map = standards_path + 'vital_unitmap.csv'
+    unit_map_df = pd.read_csv(unit_map)
+    return unit_map_df
+
+
+@pytest.fixture
+def vital_1(vital):
+    vital.loc[0, 'NVAL_NUM'] = 9999999
+    return vital
+
+
+@pytest.fixture
+def vital_2(vital):
+    vital.loc[0, 'NVAL_NUM'] = float('nan')
+    return vital
+
+
+@pytest.mark.unit
+def test_check_removal_missing_vitals(vital_1, unit_map):
+    missing_vitals = transform_vitals.filter_recorded(vital_1, unit_map)
+    assert missing_vitals.shape[0] == 0
+
+
+@pytest.mark.unit
+def test_check_removal_empty_vitals(vital_2, unit_map):
+    empty_vitals = transform_vitals.filter_recorded(vital_2, unit_map)
+    assert empty_vitals.shape[0] == 0
+
+
+@pytest.mark.unit
+def test_check_removal_unwanted_vitals(vital, unit_map):
+    vital.loc[0, 'CONCEPT_CD'] = "VISIT:PULSE"
+    empty_labs = transform_vitals.filter_recorded(vital, unit_map)
+    assert empty_labs.shape[0] == 0