The method of payment of the patient is commonly included in healthcare datasets and usually, certain payment types are associated with better or worse outcomes. Patients with no expected source of payment (NOPAY), or with Medicaid (PAYMCAID) or Medicare (PAYMCARE), typically are less healthy than patients with private insurance (PAYPRIV) or who are paying on their own (PAYSELF). Let's include all of the financial variables except for the PAYTYPER variable, which is just a nonbinary expansion of the other payment variables:

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