If you have an electronic copy (via email) then you should be able to run that through Forms to extract the data. You can even use the email endpoint if you want to automate that entire process.
As for the paper versions (mail and fax) you may want to look at adding a barcode to the document. Using Adobe LiveCycle Barcoded Forms ES you can extract the data directly.
If you can't use a barcode, my only thought is to scan the docs in and use PDF Generator to create an PDF that is OCR. The problem then is identifying the data from the rest of the text (assuming that the OCR worked well)