RMIS-Web Risk & Insurance RMIS Article

"Expert systems' savings; A large, accurate data base can reduce costs significantly"
Business Insurance, July 27, 1992

by David Tweedy

ARTIFICIAL INTELLIGENCE -- or expert systems -- may revolutionize the claims industry. In last month's column, I introduced the topic of applying this technology to claims management, particularly reserving (BI, June 22).

This computer technology has significant implications for users of claims services, be they self-administered companies, self-insured companies using third-party administrators, companies using unbundled insurance services, state funds or large governmental entities.

This month, I want to illustrate how such a knowledge-based system can reduce the cost of the claims management program to both the end user -- whether insured or self-insured -- and to the provider of those services.

Because Care Systems Corp., an Australian company in partnership with Perot Systems Corp., seems to offer the most comprehensive expert system for claims management to date, I would like to use that system for illustrative purposes.

First, the essentials: Just how does an expert claims system help to reduce claims costs?

The lifeblood of an expert system is the construction of its comprehensive data base. It is not merely a repository of historical claims information, though that is certainly an important part. Statistically significant amounts of comprehensive data on claims are broken down into many parts, such as by name, Social Security number, occupation, age and sex. In fact, the CARE system -- Compensation And Rehabilitation Expertise -- strives to capture as much data related to a claim as possible, thereby eliminating costly administrative delays that are so frequent in the claims industry.

The company's target is a sophisticated data base providing 80% of the information needed to process a claim, as opposed to 30% available in many other automated workers compensation claims systems. In fact, CARE's data base includes information on specific jurisdictions, injury codings by site and severity, special rehabilitation information, utilization patterns and cost patterns.

Certainly, constructing a sophisticated data base is a time-consuming, detailed procedure. But that effort is necessary, because the next step deals with the reserving algorithms, various claims tests independent of the data bases and the rule base, which essentially forms the basis of decision-making for the expert system.

These knowledge-based routines are what separates the CARE data base from a typical data base, regardless of size. The CARE data base can perform algorithms and other analysis to provide the continuous testing of reserve computations for each claim based on the facts of the claim as they are typed into the system.

Now, what does this mean in English? How are savings possibly derived from a more sophisticated data base?

The answer lies in the data base's accuracy. Typically, a claims examiner, adjuster or supervisor will reach a reserving decision based on facts that he or she collects, matching them against a similar claim in his or her experience. This personal philosophy is then tempered by the particular organization for which he or she works. Most large insurers tend toward conservative reserving, while TPAs vary up and down the ladder, as do self-administered organizations. Some don't reserve at all.

By using an expert system, the computer arrives at a logical valuation of the claim based on the facts of the accident, the nature of the injury, the doctor treating the individual, the region of the country in which the accident occurred, any jurisdictional peculiarity, the attorney representing the claimant, the rehabilitation specialist assigned to the case, the potential for light duty available, whether the state is a fee schedule state, whether a PPO exists, and so on.

How can a claims person possibly correlate all that data to reach a logical claim valuation over the life of a claim?

It is enough of a problem to make sure that all the facts have been gathered in a particular case (which unfortunately is something that even artificial intelligence cannot solve -- getting the claims person to complete the investigation in a timely and thorough fashion). I maintain that it is impossible to reserve for claims accurately, given the hundreds of claims that each individual claims person handles during a single year.

CARE's belief -- as well as my own -- is that claims reserving is a key claims management tool through which all claims and rehabilitation decisions should be made. To rely on consistently inaccurate reserves for claims that are outdated is not logical. The problem becomes exacerbated by the tendency individual claims people have in looking at each claim in isolation. Yet, we know that back claims have certain similarities, especially when a certain doctor or lawyer is involved in two or more individual cases. Why treat them differently?

Through the use of an expert system, however, this problem can be eliminated or greatly minimized. Decisions regarding the reserving process are made consistently in a standardized uniform format. This also assures the end user of better quality control.

Tests that expert systems perform to improve quality control in the reserving process include:

Date-based tests

Utilizing three categories of dates -- recorded dates, like the date of information receipt or disability statements; inserted dates by the claims manager or case worker; and system-inserted dates, like projected dates of partial or total incapacity -- the system automatically subjects each claim to several tests, formulating them into a date array.

This array represents a picture of each claim as of each day the claim person is working on the system in terms of past history and/or current and projected claim development. From this array, the system prompts reports to be generated when a failure to meet certain target dates occurs. This could be as simple as a claimant failing to return to work on a projected date based on the disability set by his or her own doctor.

Dollar test

The system evaluates total payments to date for each claim against the total data base, notifying the examiner whenever a payment pattern appears that contradicts a ''cost profile'' derived from the comprehensive data base.

Claim profile or rule base

This is a vital part of the knowledge-based system. It contains many conditions or ''rules'' that each claim is evaluated against, resulting in action reports to the examiner on how the particular claim compared.

The CARE system

This also includes a medical profile, rehabilitation profile, an inactivity test -- essentially no transaction with respect to an individual claim within a defined number of days.

To summarize this section, an expert system is able to provide accurate, consistent valuations for claims across a wide diversity of jurisdictions. The claims expert does not have to rely on human experience which, though extremely valuable, cannot possibly contain all of the data stored by the system.

Proactive claims management is a natural extension of accurate and consistent valuation. If a claim is consistently subjected to all of the evaluation and analysis provided by the knowledge-based system and is part of a very sophisticated data base, it would follow that much of the mundane activity of a claims person can be eliminated. That leaves time for more proactive involvement.

The CARE system provides action reports based on the tests we saw above. When a claim's profile falls into a problem area, the computer will generate an action report that suggests daily actions to be taken in the management of that claim. It is more than just a diary; it has things to do.

For example, when the system detects that more treatments have taken place than are statistically acceptable for the type of injury and the age of the particular person, it will produce an action report highlighting that problem. It will do the same for rehabilitation as well.

Now, lest anyone think that I am openly advocating an abandonment of human involvement in the claims process, let me say that the expert system simply provides suggestions or facts upon which the claims professional should act. At any time, the examiner can override the recommendations of the computer, adjusting reserves as he or she sees fit for the particular case. Yet, it does free the examiner up to do more of the thinking and planning in evaluation of claims management. The action reports from the expert system are intended to alert the claim person to certain conditions in a timely manner.

Rather than be reactive and waiting for things to happen before taking any action, the claims person can anticipate coming problems through the prompting of the expert system. He or she can, for example, intervene with an early independent medical exam or push more heavily for light duty earlier than the treating physician has recommended, based on the recommendations of the expert system.

He or she also may want to settle or commute a case faster than he or she normally would after seeing the profile of the particular claim as it is developing.

Effective utilization of manpower not only is good for the end user, but also for the claims provider. The expert system can legitimately allow for a bigger case load to claims personnel -- since it is doing a lot of the correlating and analyzing of the claims data -- and also for a smaller staff.

I am aware of one TPA that, through the use of a standard claims management information system -- not an expert system -- reduced its staff size by 30% and increased its claims handling production. It also reaped a higher profitability through this system's greater efficiency. CARE's experience in Australia is very similar to that.

To the end user, being served by a TPA or insurer utilizing an expert system means that the company should have a very efficient, cost-effective way of handling claims. This does not replace the need for claims organizations to have superior individuals handling the claims.

Claims management remains a skill that must be performed by people. Facts must be gathered, contact with the injured party must be made and interaction with medical providers, attorneys, employers and rehabilitation specialists all must occur. If, however, administrative functions can be absorbed by the system, profitability and efficiency will increase and costs should decrease.

To summarize, I believe that expert systems for claims management -- and underwriting -- are vital to the insurance industry in the 21st century. They offer an excellent opportunity for employers to reduce the cost of claims through more accurate, consistent claims handling. They offer their service providers an opportunity to become more profitable through maximization of their staffs' capability. While the expert claims system alone will not solve the rising cost of casualty claims, it is an important and necessary tool to accomplish that goal.

"Expert systems' savings; A large, accurate data base can reduce costs significantly" Business Insurance, July 27, 1992

"Expert systems' savings; A large, accurate data base can reduce costs significantly"
Business Insurance, July 27, 1992