[an error occurred while processing this directive]



Book Contents

Ch. 10
Building Knowledge-Driven DSS and Mining Data

Chapter Contents
Previous Page
Next Page

Case Study - For Underwriting, NC Blue Turns to an Expert

Joint effort with PLATINUM technology to increase accuracy and productivity

By Christy Tauhert April 2, 1998

Expert systems are quite a draw for health insurers that want to automate and improve the accuracy and efficiency of their underwriting process, especially when mistakes affect the bottom line. This was a concern at Blue Cross Blue Shield of North Carolina (BCBSNC, Durham, NC, nearly $1 billion in assets), which embarked on an underwriting expert systems project for medical applications with PLATINUM technology (Oakbrook Terrace, IL) that not only increased the accuracy of underwriting decisions, but cut the underwriting process in its small group health unit from five days to just minutes.

"We saw the need to get more consistency in our underwriting decisions and increase productivity in the underwriting process," says John Friesen, vice president of BCBSNC's actuarial and underwriting services unit. The NC Blue, which offers managed care, traditional indemnity, group and individual health insurance, had such concerns in mind when it came across PLATINUM in mid-1995. PLATINUM's programming expertise, combined with BCBSNC's underwriting expertise, proved to be a good fit for a joint project to develop an automated underwriting system, says Friesen.

PLATINUM consultants worked with the insurer's underwriting and IT employees to co-develop an automated underwriting expert system framework using PLATINUM's Aion application development environment. Setting up the core system involved gathering and creating rules for the system. Business rules were created according to 157 ICD-9 codes (International Classification of Diseases medical condition codes). The automated system asks for the patient's conditions, medications and treatments. When it has enough information, the system makes a decision accordingly, says Friesen.

Once rules had been created in the Aion rule base, a Powerbuilder graphical user interface was built to collect enrollment data on Windows 95-based client PCs (used by 10 to 20 underwriters) and pass it to Aion. They then process the data and feed the results to an IMS database server that is supported by an MVS-based IBM mainframe. Sybase NetGateway, a database interface, integrates the rules base in Aion with the IMS database, client PCs, a Unix server, the mainframe and BCBSNC's proprietary rate quoting, risk management and policy writing systems. A second database, Oracle (running a Unix server), receives enrollment data for management reporting.

For the pilot, BCBSNC used a few medical conditions to make sure the underwriting decisions were being made accurately. A complete implementation process began in April 1997 to create the rest of the rules, and the system went live in September 1997.

The solution has saved the NC Blue money in terms of the type of employees it needs to facilitate the underwriting process, which now only requires analysts or enrollment entry people to enter key information into the system. As a result, BCBSNC was able to eliminate two full-time employees and change responsibilities for three others.

The solution also saves time. "Before, the process took anywhere from one to five days to complete. Now, it takes just as much time as it does to enter information into the computer it's a matter of minutes," notes Friesen.

BCBSNC's goal was to use the system for 60 to 70 percent of its applications, although the number has grown to 85 percent. Friesen says he's hesitant to take that number higher, because "there are complex medical conditions I wouldn't want the machine to handle."

While Friesen would not discuss costs, return on investment will take less than a year, because accuracy of underwriting decisions has been improved by well more than one percent, he says.

Reprinted from the March 1998 issue of Insurance & Technology magazine. Copyright © 1998 Miller Freeman, Inc., a United News & Media Company, 600 Harrison Street, San Francisco, CA 94107, 1-415-905-2200

Questions for Discussion:

    1. Is this underwriting expert system a Knowledge-Driven DSS? Why or why not?
    2. What are the benefits of this expert system?
    3. Why is the expert system not used to make all underwriting decisions?
    4. What is the architecture for the system? Draw a flow chart.
[an error occurred while processing this directive]