Blue Cross And Blue Shield's Imaging Upgrade
A new scanning platform helped the $3.8 billion health insurer improve productivity by reducing rescans more than 85%.
Many of the claims Blue Cross and Blue Shield of North Carolina receives include both a claim form and a check payment. To ensure accuracy, it is imperative that these images be kept together. Traditionally, grouping these images was not a hardware function. Either images would have to be manually reassociated with one another postscan, or custom software could be developed that would automate this process for you. However, the ImageTrac III not only handles documents of various sizes and densities, it maintains transaction integrity by linking the form and accompanying check images scanned from a single envelope. This capability eliminates much of the back end work that used to be necessary to keep a transaction intact, allowing your employees and your imaging system to be more productive.
Imaging Business Machines LLC’s (IBML’s) ImageTrac III scans 120 to 140 ppm (pages per minute) in color, monochrome, or grayscale, and its open track design allows for visibility and access of documents throughout the scan process. This helps users quickly recover from document jams and provides a high tolerance for creased documents, staples, or paper clips. The ImageTrac III also allows up to three pockets to be added to the scanner to aid in postscan document collection. Real-time pocketing decisions can be based off intelligent features of the cameras, including in-line ICR (intelligent document recognition) and in-line document classification, eliminating manual resorting of the document sequence.
For More Info. On Imaging Business Machines LLC
Go To www.ibml.com
Automatic Classification Reduces Document Prep Labor
As mentioned in the main body of the story, Blue Cross and Blue Shield of North Carolina is currently considering adding automatic document classification technology to its scanning platform. This data recognition software eliminates the need for the manual sorting of documents and insertion of separator sheets prior to the scanning process. This is made possible by the self-learning capabilities of the software. Most software suites in this category examine the text and layout characteristics of each document and match these characteristics to a common document type (e.g. invoice, sales order, professional claim form, institutional claim form). These technologies take a probabilistic approach in which the key attributes of each document type are automatically determined from a set of examples, instead of deciding which words and phrases are important and manually encoding these as rules. These key attributes are used to generate a signature for each document type. When the software encounters a new document, a probability is assigned indicating how well that document matches each of the prelearnt types. The software not only designates the most likely document type, but also provides a measure of confidence in that assessment. This allows low-confidence classifications to be reviewed by an operator to minimize errors, similar to the way in which questionable individual characters are reviewed and validated when processing fixed-form data. Once a document is classified correctly, it can also be automatically routed to the next step in its workflow. For example, different claim types can be sent to the appropriate departments for further processing.
The ability to automatically classify documents can drastically reduce your manual document preparation expenses and accelerate your document imaging processes. However, while these technologies have come a long way over the past year, there are still some limitations of the technology that you should be aware of. First, teaching the software to learn the different (and sometimes custom) document types your organization must process can be a time-consuming process that involves a lot of tweaking. Be p
While all high-volume document imaging platforms are designed to eliminate much of the manual labor involved in paper-intensive forms processing environments, some do the job better than others. For example, Blue Cross and Blue Shield of North Carolina’s (BCBSNC’s) image and distribution center (IDC) initially implemented a scanning platform in 1998 that required quite a bit of manual intervention to image the 180,000 pieces of paper the insurer receives each day.
The system consisted of seven high-volume document scanners, each of which required one specially trained full-time employee to operate. Twenty-one full-time employees were also required to perform document preparation tasks, such as physically sorting documents into different batches based on document type, removing staples and paper clips from each document, and inserting bar coded separator sheets to alert the system to a batch or template change. Much of the document preparation labor was actually a direct result of an inherent weakness in the scanning devices themselves. The scanners BCBSNC initially invested in were only equipped to handle documents 8½ by 11 inches in size. However, BCBSNC needed to image documents of all sizes, including checks, prescription forms, and receipts. In order to process these irregular documents, BCBSNC employees had to tape these items to a blank 8½-by-11-inch sheet of paper prior to scanning. These paper-handling limitations, combined with questionable image quality, also resulted in 9% of the total documents imaged needing to be rescanned.
Even with these limitations, BCBSNC’s existing imaging platform sufficiently handled the normal daily volume of 180,000 claim and other documents. However, once the core benefits of imaging were realized in IDC (e.g. fast and simultaneous access to documents, reduction in storage space requirements, electronic routing, etc.), other departments in the organization wanted to leverage the scanning system to image and archive their documents electronically. Initially, it was thought that these new requests could be accommodated after hours, but the scanning platform was dependent on BCBSNC’s operating system, which was taken offline between 9 p.m. and 6 a.m. each day so data backups and other IT maintenance procedures could be performed. Unable to satisfy growing internal imaging demands with its current platform, BCBSNC launched an initiative to develop a new scanning infrastructure that could support the volumes and variety of paper from multiple departments and also drive down processing costs through centralized efficiency.
IMPROVED SCANNING SPEEDS CUT HARDWARE REQUIREMENTS IN HALF
BCBSNC began evaluating several new high-volume scanning models in late 2004. In particular, the insurer sought a scanning platform that was faster, could handle documents of varying sizes and densities, had optimal image quality, and could integrate with BCBSNC’s EMC/Captiva FormWare processing platform. Based on these criteria, BCBSNC opted for ImageTrac III scanners from Imaging Business Machines LLC (IBML).
The scanning speed alone of the ImageTrac made an immediate impact. “Productivity measurements of our old imaging platform showed that each scanner in this system processed an average of 52 ppm [pages per minute],” says Christy Lane, operations manager of the imaging and distribution center for BCBSNC. “A demonstration by IBML using our documents proved that we could process an average of 130 ppm on each scanner by moving to the ImageTrac platform. This allowed us to cut our hardware requirements by more than half.”
BCBSNC purchased three IBML ImageTrac III scanners to replace its seven previous models. This move not only allowed the insurer to produce more work with fewer devices, it helped BCBSNC’s IDC eliminate labor costs because now only three scanner operators were required instead of seven.
BCBSNC knew the implementation, integration, and switch over to the new IBML scanning platform would take a few months for its IT department to complete, but the IDC arranged to have the scanners onsite for training and testing purposes long before the system went live. “Working with the ImageTracs prior to the system becoming operational allowed our scanner operators to familiarize themselves with the new hardware,” says Lane. “It also allowed us to recognize and remedy any kinks we were likely to encounter before the scanners were actually used in a production environment.”
ENSURE NEW IMAGES FIT EXISTING TEMPLATES
BCBSNC encountered two specific kinks during the testing phase. First, while impressed with the new scanning technology, BCBSNC’s scanning operators had a hard time breaking themselves of their old scanning habits, which initially limited the capabilities of the ImageTrac hardware. For example, under the old imaging system, scanner operators were used to placing one batch of claims in the document feeder at a time and waiting for that batch to run before inserting another set of documents. They continued scanning batches one at a time on the IBML platform, even though the ImageTrac is capable of holding multiple batches in its document feeder and scanning these batches back to back. BCBSNC’s IDC management team had to physically stand with scanner operators at the device to break them of this habit and ensure they took full advantage of the ImageTrac design.
The other kink BCBSNC encountered was more technical, in that it had to retrain its FormWare system to recognize the new images that were being produced by the ImageTrac scanners. Various scanning models line up images differently. The ImageTrac scanners crop an image the moment a document is scanned. Image cropping was not a function performed by BCBSNC’s previous scanners and therefore was performed postscan by the FormWare software. BCBSNC had to adjust the settings in FormWare to ensure the images produced by the ImageTrac scanners lined up with existing processing templates for OCR (optical character recognition) and data extraction purposes.
REDUCE RESCANS WITH IMPROVED PAPER HANDLING, IMAGE QUALITY
BCBSNC fully launched its IBML scanning platform in mid-2005, and speed is not the only way the new system has proven more efficient. The ImageTrac’s open track design allows documents of any size to be scanned, eliminating the need for BCBSNC employees to tape irregular documents to 8½-by-11-inch sheets of paper. This labor reduction now allows document preparation work to be performed by 19 employees rather than 21.
“The improved paper handling of our new imaging system also results in fewer paper jams,” says Lane. “This, combined with the improved image quality we now receive, has allowed us to reduce instances of rescans from 9% of our total volume to 1.3%. This is a reduction of more than 85%.”
Under its new imaging system, BCBSNC’s IDC receives mail from the mailroom where envelopes are opened and documents are prepped for scanning. Contents are separated and sorted by document type such as claims and non-claims. A bar-coded separator sheet is inserted to alert the system to the type of document contained in the batch so it can match the images to the correct template and route them to the appropriate internal processing group. After scanning, images are routed to FormWare where OCR is performed on specified forms and human processors verify data unrecognized by the software. Processors may also key additional information contained on the image into FormWare. Once these steps are complete, a set of custom business rules are run on each image based on document type. For example, the system will automatically compare the member number or diagnostic code listed on the claim to the ones listed in the BCBSNC database to ensure they are valid. From there, images are routed to data entry operators where they are manually indexed based on several metadata, including document type, member number, member name, and provider number. These images are then stored in a FileNet content management system where they can be quickly accessed via PC by BCBSNC customer service representatives.
BCBSNC’s new imaging system not only helped the insurer streamline efficiencies in the claims department, it allowed the platform to be leveraged to accommodate other imaging projects throughout the organization, such as scanning and archiving member enrollment information. This is possible because the ImageTracs come equipped with their own database that allows them to function independently of the BCBSNC operating system. The ImageTracs can run 24/7 if needed, and images captured after hours can simply be transferred from the ImageTrac database to the BCBSNC database once the next workday begins.
BCBSNC also hopes to further streamline its claims processing procedures by integrating unstructured data recognition and automatic document classification technology with the platform in the near future. These technologies allow different document types to be identified and processed without the need for presorting and separator sheet insertion. IBML is currently bundling these technologies with its ImageTrac scanning platform and plans to pilot these solutions this summer.