Calculate look back period using pandas

Question

I need to look back backwards in each row for a different time and count new patients, so its like I look back from a date and check new patients in last month then check new patients in second last month and so on.

My data set is something like this: In this image, I want to start from last RX_FILL_DT and then check from last month to two years This is my Data Set

This is my Data Set

PATIENT_Id > RX_FILL_DT > RX_QUANTITY > DAYS_SUPPLY

106208 > 3/25/2010 > 20 > 30

1240316 > 5/15/2010 > 15 > 30

1381727 > 8/26/2010 > 10 > 28

1381727 > 1/17/2010 > 10 > 28

1381808 > 6/6/2010 > 30 > 30

1381938 > 4/16/2010 > 45 > 90

1381938 > 8/3/2010 > 15 > 30

1381938 > 9/4/2010 > 15 > 30

1381938 > 9/14/2010 > 15 > 30

1381938 > 9/30/2010 > 15 > 27

1381961 > 6/8/2010 > 15 > 25

1381961 > 7/1/2010 > 15 > 23

1381961 > 7/8/2010 > 15 > 25

1381961 > 7/28/2010 > 15 > 23

1381961 > 8/19/2010 > 15 > 25

Max Date 9/30/2010

And my output should look like this:

             lookback 1 month >   lookback 2 month    >lookback 3 months

No of new patients > 0 > 0 > 0
Lookback calculation can been seen here

New Patients = post - common (between prior and post)

Here the patient is new if they don't have any record in look back time frame.

For eg, for a look back of one month, a particular patient id has only one entry then it means the patient is new and is counted when lookback is one month.

Please help I am new to pandas.

My code is like this:

onlyDip = pd.read_table("C:\Users\aa18957\Desktop\Transactions.txt", sep = "|" )

onlyDip['RX_FILL_DT'] = pd.to_datetime(onlyDip['RX_FILL_DT'])

I am trying to use pandas.Series.rolling but I am not sure how to make it count new patients.