Medical tests, while important, occur post-prediction time and must be omitted for this use case. They may be used for other modeling tasks, such as readmission prediction or mortality:

testing_cols = [
    'ABG','BAC','BLOODCX','BNP','BUNCREAT',
    'CARDENZ','CBC','DDIMER','ELECTROL','GLUCOSE',
    'LACTATE','LFT','PTTINR','OTHERBLD','CARDMON',
    'EKG','HIVTEST','FLUTEST','PREGTEST','TOXSCREN',
    'URINE','WOUNDCX','URINECX','OTHRTEST','ANYIMAGE',
    'XRAY','IVCONTRAST','CATSCAN','CTAB','CTCHEST',
    'CTHEAD','CTOTHER','CTUNK','MRI','ULTRASND',
    'OTHIMAGE','TOTDIAG','DIAGSCRN'
]

X_train.drop(testing_cols, axis=1, inplace=True)
X_test.drop(testing_cols, axis=1, inplace=True)