| Home | Software Providers | Consultants | Articles | Columns | Reviews | Headlines |

-

Entire contents Copyright © 1999 Business Insurance

"Artificial intelligence; Knowledge-based comp claims system ready for U.S. market"
Business Insurance, June 22, 1992

by David Tweedy

APPLYING ARTIFICIAL intelligence -- or expert systems -- to the claims reserving function within the insurance industry is overdue.

In my Jan. 20 column on artificial intelligence in underwriting systems, I asked why the insurance industry had not more actively embraced expert systems in the two most prevalent and logical functions within insurance: underwriting and claims.

As a former claims adjuster/supervisor for Liberty Mutual Insurance Co., my initial reaction to artificial intelligence and case reserving was predictable and generally reflective of the claims industry's opinion: ''Reserving is an art, not a science....Computers will never be able to include all of the objective and subjective elements that go into setting future reserves in workers compensation or liability cases.''

Unfortunately, neither does the claims industry -- there is so much variance on how to set casualty reserves on a case basis within the industry itself! However, no one in the insurance industry seemed to be interested in providing a sophisticated reserving system to improve and streamline the process.

That has changed. In 1991, Risk Data Corp. introduced Micro Insurance Reserve Analysis to the insurance industry. MIRA provides on-line automatic reserve setting, drawing on a historical data base of 10 million claims from 14 insurers.

And now, Care Systems Corp. -- an Australian company with offices in Dallas -- in partnership with Perot Systems Corp. of Dallas is unveiling a comprehensive workers compensation claims management information system that includes computer-derived reserves. (Until recently, Perot Systems was headed by Ross Perot.) The system -- Compensation And Rehabilitation Expertise, or CARE -- is being introduced to the U.S. and Canadian workers comp market, hoping to parallel or even improve upon its performance in Australia.

Over a 3 1/2-year trial period utilizing the CARE system, average claims costs in Australia were reduced to $ 384 from $ 2,300. Whereas the average industry loss ratio in Australia was 80%, Australian firms using the CARE system reduced the loss ratio of claims to premiums to 30%. Furthermore, CARE was able to reduce internal administration costs by more than 50%: For every $ 10 million of premium under the old system, 10 people would be required to maintain it. With the CARE system, that was cut to three.

Obviously, Care/Perot is looking to reach the same level of success in North America -- the largest workers compensation marketplace in the world.

How can this system have results like this? Can it be transferred to the highly complex, combative and beleaguered U.S. workers comp system? Just how does a knowledge-based system reduce health claims and administrative costs? What about other emerging expert systems? Finally, how does CARE fit within the risk management information system industry?

These are some of the same questions that we were asking when Perot Systems asked us to review the CARE system last December. After all, entering a mature marketplace like the RMIS industry requires the product to clearly differentiate itself from its competition. Otherwise, it would get lost in the crowd.

Clearly, the advantages of a knowledge-based claims information system make it unique in comparison to any of the mainframe, minicomputer or microcomputer vendors available today. Many are quite sophisticated, using fourth-generation languages, image processing, computer-aided software engineering tools, and some quasi-artificial intelligence, especially in the more sophisticated case management modules on medical treatment. None, though, has designed a core system for claims management as a knowledge-based system.

For illustrative purposes, I'd like to discuss just how CARE's knowledge-based claims management system works from an overview perspective:

Comprehensive data base

For any knowledge-based system, the construction of a comprehensive and detailed data base forms a crucial foundation. In fact, a sophisticated data base provides 80% of all the information needed to process a claim, whereas, according to Care, other workers comp systems only provide 30% of the necessary information.

The data base itself is a compilation of dozens of data elements crucial to the workers comp process: litigation, including comparative verdicts in different jurisdictions; an incapacity data base, including periods of hospitalization, total disability and partial disability; a medical data base, including the customary injury codes and classifications; a rehabilitation data base; and standard demographic information like occupation, age and sex. The MIRA system, for example, is able to analyze up to 66 individual claim characteristics.

In essence, necessary claims characteristics are captured as data and input into the master data base upon which the rest of the knowledge-based system's features depend.

Advanced claims management capabilities

A comprehensive, knowledge-based system, like CARE, has the following similarities to other advanced systems:

• Automatic processing of claims, minimizing manual input.
• Reduction of paper file dependence.
• Effective quality control of claims-managed decision.

Most RMIS workers comp systems can do these things.

Two distinctive features of the knowledge-based systems are automatic reserving and pro-active management reports.

Both the CARE and MIRA systems utilize an extensive data base to compute indemnity, medical, rehabilitation and expense reserves on incoming claims.

Using certain proprietary rule bases -- a major foundation of a knowledge-based system -- the CARE system evaluates each new claim, subjecting it to a number of complex tests to produce the reserve estimate.

How accurate is it? MIRA's accuracy has been computed at 98%, validated by Ernst & Young, based on tests performed at two leading insurers. CARE's accuracy rate is similar, based on its experience in Australia.

The obvious advantage is in saving time. If claims information is gathered and entered quickly, the system can generate an accurate reserve without some of the guesswork frequenting many claims organizations. The suggested reserve is not mandatory; the claim examiner has veto capability if he or she knows of some fact that has not been made part of the data base or the claim itself. The suggested reserves are guidelines and timesavers only.

Further, these reserves are constantly evaluated and changed over time. Obviously, the outstanding reserve is decreased as the claim advances in time and approaches its ultimate result. However, additional information as to medical, legal or liability status would affect the reserve as well, either up or down. It is, therefore, a dynamic reserving process.

Another advantage of the knowledge-based system is its creation of ''action reports.''

Given daily to claims personnel, these reports comprise a series of tests for all workers compensation claims in the CARE data base. An action report is produced for every claim identified by the CARE system as requiring reaction or other followup by claims personnel. It is not just a simple diary function. Each action report displays key data, like an accident/claims summary, an injury summary, current claims status, current reserve estimates and particular instructions to individual claims personnel for followup.

The action reports are a supplemental tool for the claims professional; they are not meant to supplant the human processes of devising tactics and strategy in investigation and resolution of a claim. They are intended to alert them beyond just a simple diary function, since the system can anticipate conditions and key action areas. A diary system would simply print out, upon instruction, a list of claims to be reviewed every 20 or 30 days. The CARE action report would take basic information contained within the claim itself and identify action steps to be taken at any time.

In next month's column, I will examine why a knowledge-based system can reduce the cost of the claims program to both the end user, whether self-insured or insured, and the provider of those services.

Copyright© 1992 Business Insurance