--- a
+++ b/analysis/read_file.py
@@ -0,0 +1,51 @@
+import pandas as pd
+import numpy as np
+import pdb
+
+def read_file(filename, longitudinal=False, rare=True):
+    """read in EHR data."""
+    xd_name = filename + '_demographics.csv'
+    if not longitudinal:
+        xc_name = filename + '_common_median_imputed.csv'
+    xr_name = filename + '_rare.csv'
+    label_name = filename + '_class.csv'
+
+    xd = pd.read_csv(xd_name,index_col='PT_ID')
+    if not longitudinal: 
+        xc = pd.read_csv(xc_name,index_col='PT_ID')
+    if rare:
+        xr = pd.read_csv(xr_name,index_col='PT_ID')
+    
+    label = pd.read_csv(label_name,index_col='PT_ID')
+    
+    print('longitudinal =',longitudinal,'rare =',rare)
+
+    if not longitudinal and rare:   # demographics, common, and rare labs
+        df_X = pd.concat([xd, xc, xr],axis=1)
+        print('loading demographics, common, and rare labs')
+    elif not longitudinal:  # keep common labs in there, remove rare
+        df_X = pd.concat([xd, xc],axis=1)
+        print('loading demographics and common labs (rare = ',rare,')')
+    elif not rare:  # if longitudinal AND don't include rare, use only demographics
+        df_X = xd
+        print('loading demographics only (longitudinal = ',longitudinal,')')
+    else:   # for longitudinal case with rare, remove common labs, include everything else
+        df_X = pd.concat([xd, xr],axis=1)
+        print('loading demographics and rare labs (longitudinal = ',longitudinal,')')
+    
+    assert(all(df_X.index==label.index))
+    ###
+    # Drop total cholesterol (sorry for the hack)
+    if '2093-3' in df_X.columns:
+        print('dropping total cholesterol')
+        df_X = df_X.drop('2093-3',axis=1)
+
+    feature_names = np.array([x for x in df_X.columns.values if x != 'class'])
+
+    X = df_X.values #.astype(float)
+    y = label.values.flatten()
+    pt_ids = df_X.index.values
+
+    assert(X.shape[1] == feature_names.shape[0])
+    # pdb.set_trace()
+    return X, y, pt_ids, feature_names, filename + '_long_imputed.csv'