KodeNet works for Medical Insurance companies/ TPA converting the paper claims into digital format.

KodeNet processes Hcfa 1500, UB92, ADA & EOMB (Explanation of Medical Benefit) forms daily by scanning/imaging these documents or accepting them from our clients via ftp and either OCR or manual double key & verify process. The use of our key-from image technology will allow us to route non-readable (ocr) forms to this system where we will double key from images any unrecognized data. The output data is formatted as electronic claims which are transmitted to our clients in an ascii comma delimited, flat ascii, Hipaa 837 X12, 835_4010 or NSF format (or any other format they might require) for processing.

We can process individual HCFA and institutional UB92 claims as well as dental claims. In the mailroom, the claims get opened, removed from their envelopes and sorted into separate piles: single sheet claims, multiple sheet claims, HCFA's, UB92's, Dental Claims, and miscellaneous other sheets which cannot be processed and quality level of each form. We will not process a document which does not contain a claim form, therefore, any claim or correspondence, whether single or multi-sheet, must always contain a claim form.

The forms are scanned at 200-300 dpi depending on the quality of the form coming into process. During the time pages are scanned the fields are OCR'd and validated for field content. If acceptable, a control number is assigned. The pages are then imaged in a tif format and batched. While scanning we use a red filter located in the scanner to remove the form. As a result, our images are approximately 15k or so each. Critical to our process are two exception handlers. This exception handler allows the system to determine if the claims are good enough to OCR ( any claims that do not meet validation rules or that do not OCR at more than 80% accuracy go to this exception) if they are not the claims is re-routed to our exception program where the claims will be keyed from image.

Well we found that it is more effective for an operator to key all the data if more than 20% needs to be corrected. If there is more than one character wrong in a field it is quicker for our operators to re-key the whole field than try to fix it. Statistically, a more than 20% error rate means that most fields will contain more than one error.

Once the forms are scanned and OCR'd successfully they get assigned a control number and an image number which is indexed in the output file associated with that claim. The images get stored on our Windows NT Server where we will keep them their for our clients to access them through our online search or where they will be transferred to a CD-ROM or Drive of their choice. The rejected pages are either re-scanned as images or re-routed to our exception key entry operators. While this is going on, our key entry stations are looking for batches of images that need to be corrected or key entered. The batch is opened on a verification station creating a split screen displaying the image on the right and the zoomed in zone on the left and key entry fields below with any unreadable characters displayed by highlighting it in yellow.

Once the batch is complete, and the image record is flagged as keyed, the automated job-flow process sends the batch through the output queue into a text file holding the job in an ascii comma delimited format. At the end of the day we generate a transmission report with the file which details each of the batches which were processed.