Add in old radiation history data.

scripts/transform_radiation_history.py

@@ -25,6 +25,10 @@ def get_args():
                         help="Radiation History File (csv format).",
                         required=True)
 
+    parser.add_argument('-a', '--additional',
+                        help="Radiation History addtiional File (csv format).",
+                        required=False)
+
     parser.add_argument('-s', '--sites',
                         help="Site Map File (csv format).",
                         required=True)
@@ -134,11 +138,12 @@ def calculate_shift(radiation, date_shift):
     # Int MRN
     merged.MRN = merged.MRN.astype(int)
 
-    #TODO: Add radiation units
+    merged['dose_units'] = 'cGy'
 
     # Rename columns
     radiation_columns = ["mrn", "site_specific", "site_general",
-                   "dose", "fractions", 'start_date', 'end_date', 'treatment_intention']
+                   "dose", "fractions", 'start_date', 'end_date',
+                   'treatment_intention', "dose_units"]
     merged.columns = radiation_columns
 
     return merged
@@ -147,6 +152,7 @@ def calculate_shift(radiation, date_shift):
 def main():
     args = get_args()
     rthistory = args.file
+    additional_rthistory = args.additional
     sites = args.sites
     icd = args.icd
     date = args.date
@@ -158,6 +164,7 @@ def main():
 
     # Read in files
     radiation = pd.read_csv(rthistory)
+
     site_map = pd.read_csv(sites)
     icd_filter = pd.read_csv(icd, sep='\t')
     date_shift = pd.read_csv(date)
@@ -171,6 +178,14 @@ def main():
     # Map Specific Sites to General Sites
     fix_sites, unmapped_sites = convert_sites(fix_icd, site_map)
 
+    # Merge sites with additional data
+    if additional_rthistory:
+        radiation_add = pd.read_csv(additional_rthistory)
+        radiation_add.drop(['DIAGNOSIS'], inplace=True, axis=1)
+        merge_sites = pd.concat([fix_sites, radiation_add])
+        fix_sites = merge_sites
+
+
     # Convert missing Treatment
     fix_treatment = convert_treatment(fix_sites)