www.RiskNoodle.com

Reducing The Workload

The runaway costs of processing workers' comp insurance is a big reason cariers must invest in technology to improve workflow, classify claims and identify fraud

By Steve Dwyer, Senior Editor

March 1, 2004 - For years, administration of workers' compensation insurance at Boston-based OneBeacon Insurance Group was viewed as an occupational hazard-a bona fide accident waiting to happen.

Text Mining for Fraudulent Patterns

Using an approach of text mining of claim diaries, Magnify introduces a new product called, FraudFocus. 

FraudFocus^TM, Magnify's predictive modeling-based claims fraud product, continuously reviews claims, beginning at first notice of loss and thereafter whenever a claim is updated throughout the entire life of the claim. The "heart" of the FraudFocus^TM approach is predictive modeling, which relies on advanced statistical analysis of claims data to develop algorithms (models) that can accurately identify claims likely to be fraudulent. Predictive modeling is a proven fraud detection technology. In the credit card industry, predictive modeling has been utilized for years to reduce fraud losses by 30% or more. Opportunities exist in Worker's Comp to apply predictive models in a similar manner to identify claimant fraud, provider fraud and policy fraud simultaneously.

Couple this approach with a streaming feed of medical bill analysis, and the possibilities are endless.

What goes around...comes around!

What goes around...comes around! Well, it was bound to happen and I just couldn't resist showcasing the new IPAQ.

In the event you didn't remember:

The abacus as we know it today, appeared circa 1200 A.D. in China; in Chinese, it is called suan-pan. On each rod, this classic Chinese abacus has 2 beads on the upper deck and 5 on the lower deck; such an abacus is also referred to as a 2/5 abacus. The 2/5 style survived unchanged until about 1850 at which time the 1/5 (one bead on the top deck and five beads on the bottom deck) abacus appeared.

So, next I would expect a IPAQ Slide Rule.

Sell Once: POS and Modeling

According to Fair Issac:

The Hartford Insurance Group is gradually moving all of its decision logic onto Blaze Advisor, the world's leading business rules management solution.  Blaze Advisor enables The Hartford to combine its own predicitve models with rules and other dcisioning elements - all without programming - and rapidly deploy the automated decisioning processes to multple frontline interfaces, including third party agency management applications (with automatic transpormation from/to XML) and a cusotmer self-serve website.  Currently, new BRM-based processes are being used for real-time POS quotations and new business; soon they'll be extended to renewals as well.

Computer grids speed Predictive Modeling Results

Computer grids speed Predictive Modeling Results - Nationwide Insurance, with the assistance of 5-year old company, DataSynapse has developed a grid computing scheme to dramatically reduce time for problem solving.  A typical predictive modeling solution would take the Nationwide Actuarial department about 83 days to compute on one of its processors was now reduced to 81 minutes using a 50-computer grid.

"One source of competitive advantage is how well we analyze that data and what you can do with it.  We did expect it to succeed as well as it has.  It challenges your fundamental notions..."

Amica Mutual Develops Models for Marketing Dollars

Amica Mutual Develops Models for Marketing Dollars - After procuring senior management buy-in, Amica Mutual was able to develop an Inquiry Model that identified 12 predictive variables which were most likely to predict an inquiry into purchasing life insurance.  By segmetation and scoring, it was discovered that a vital few concepts exists in life insurance purchases.  Amica exceeded its inquiry goals by 110%, reduced marketing costs by 25% and increased revenues.

Data Mining Resources

Data mining resources can be found at Dorian Pyles website: Model & Mine.  Of particular interest is the breakdown of basics of mining and modeling. As discussed in his book, Business Modeling and Data Mining, Pyle explains just what models are and how they can be used to reveal hidden assumptions and needs, determine problems and discover data and costs.

Additionally, the site compiles a review of data mining and business modeling software such as Clementine, JMP, Brainmaker, @Risk, CrystalBall and many more.  It is a terrific thumbnail sketch of strengths of each program.

Be sure to Bookmark this page!

MAT: A new Tool for Life Insurance Underwriting

In this paper presented by BioSignia, Inc., a study is reported that demonstrates yielding a higher quality of preferred class risks when utilizing Internet-based marekting.

The evolution of the life insurance industry is requiring a robust risk assessment tool that not only could streamline the underwriting process and reduce the processing cost, but also improve the risk assessment quality. The Mortality Assessment Technology (MAT) introduced in this paper was developed for that purpose. MAT is based on multiple evidence-based disease prediction models and a conversion to predicated mortality ratio (PMR). By applying MAT into a national longitudinal cohort dataset, it was shown that MAT accurately predicted mortality and yielded better mortality risk discrimination power then the conventional method. More specifically, the preferred class defined by MAT had significantly lower mortality outcomes than the preferred class defined by conventionally used criteria. It is believed that MAT could play significant roles in the evolution of underwriting from a paper driven process into an automatic, computerized, intelligent decision process.

Need $550,000 ???

Put your mind to work and develop a new algorithm to help identify terrorist chat rooms.  RPI is looking for proposals to develop this technology as they believe as many other sources do, that the terrorists who planned September 11th used the basic email and chat rooms for their work.  In fact, this author heard of a Tampa librarian who, inadvertently believes she set up an email account for Atta.  Remember, public libraries continue to provide access to the internet for free.

Apparently, those who do plotting of any kind, and use the internet, deviate from customary practices by sending flurries of messages to only one person, for example.  It is through this type of "sorting" and mining that RPI hopes to identify these patterns and activities.  The US govt has for years relied on a 30 year old software called, Echelon to review massive amounts of phone data.

Maybe George Orwell's time is upon us!

Please, don't check my driving record!

What if liability premiums were based not on demographics and DMV records but on actual driving habits? That's the idea behind Norwich Union's newest business. In a pilot program launched last year, the British insurer began to track 5,000 customers by placing GPS receivers in the trunks of their cars. Data from the devices will be used to adjust drivers' premiums from month to month. "We calculate everything from speed and acceleration to whether they're braking too early or too late at intersections," says project manager Robert Ledger. So far Norwich Union has collected data on about 1 million journeys, and its statisticians are using the information to recalculate its insurance tables. Next year it will make the technology available to the rest of its 3.5 million auto customers, hoping that the feedback will lead to fewer accidents -- and thus higher profits. "Over time," Ledger says, "we think we can change people's behavior and make them safer drivers."

Or check out the actual features at Norwich Union:

Features of the Pay As You Drive’™ For Young Drivers product include:

• a tariff per mile, with the rate personalised to each customer and dependent on the time of day the car is driven
• the first 100 off peak miles each month are free
• young drivers could see a saving of up to 30% a year off the cost of their premiums
• premiums are billed monthly – premiums will vary monthly if customers vary their driving
• there is a one-off fee of £199 for the telematic device – this is expected to cost less than the savings to be made in premiums.
• the box has a built in car locator – enabling customers to have their car located if there’s been an accident, breakdown or if it has been stolen

Now, could this data collected be indicators of future success in say, bill payments, return-to-work predictors, or fradulent actions?

Ask and ye shall receive!

Ask and ye shall receive!  Not exactly biblical in nature, but Motorola will produce a mobile phone, licensed with Apple I-Pod technology.  (maybe..Talkn'Tunes...you heard it hear first).   A pretty safe prediction is that we will see a mobile phone with a 3.1 mega-pixel camera, Palm-powered applications with built-in QWERTY keypads and hi-fidelity speakers and hi-res screens to download movies and TV shows within three years.  And all for $250 and a two year contract.  Now, that's biblical!

More Incoming Data Streams

More incoming data streams...this time from eylion.  It has amassed a database of 23 million people and 1.5 million companies.  All of the data is for sale at $6.50 per name.  Check the accuracy of your name....you'll be surprised.

The data interpretation is based upon three different technologies.  They are: Natural Language Extraction (the ability to read an English sentence and know what it means), Artificial Intelligence Algorithms (providing the ability to interpret information from various websites) and finally, Information Integration Logic (an organizer and validator of the data found).

Perhaps this information could be utilized in claims management, or life and disability underwriting.  It seems like a good start, but much more data will need to be collected.  How can we get into the social security disability system or state unemployment data?  That could really provide insight into the vailidity of claims or the worthiness of a life underwriting process.

I-Pods are more than...

I-Pods are more than just an mp3 player.  Truly, with its revolutionary 1" hard drive it becomes a mobile data storage device.  Since, the new I-Pod technology includes the ability for color pictures, many people are turning them into handle photo albums.

But here's an innovative example of using an I-Pod in medical setting.  Radiologists are turning to I-Pods to display scan images. While resolution is not high enough for accurate reading, it has become an interesting tool for reaching out to patients and helping in their education of the results.

The special software called, " OsiriX" can be downloaded FREE from its home page.  How long can we wait for Motorola and Apple to team up on an I-Pod phone?

e-Prescriptions are making the rounds!

e-Prescriptions, using a handheld device to capture and electronically deliver medical information are hitting the mainstream at Boston hospitals.  Today, the Boston Globe reported that medical plans are now encouraging the use of the handhelds.  Some benefits include:

Health insurers are interested in paying for e-prescribing partly because the software checks whether the doctor has chosen the cheapest drug, and if not, suggests a lower-cost alternative, which could save insurers money on soaring drug costs.

Insurers also say e-prescribing will reduce the number of patients who get the wrong medicine or dose because the pharmacist can't read a doctor's handwriting.

In addition, doctors on-call can use the handheld to immediately check on a patients drug interaction as well as reduce errors in transcription and speed up the care and treatment of the patient.

Two companies are marketing the handheld, Zix Corp. (www.zixcorp.com) and DrFirst (www.drfirst.com).  Both companies offer a full-suite approach to the handheld tool, ie. medical records, appointment scheduling, billing etc.  The only advice: Just keep it dry!

HP Mobile Agent is Key Solution for Insurance Agents

Your worst nightmare is about to come true.  Imagine thousands of life insurance agents armed with handheld devices and capturing information about you for an on-the-spot life quote.  Or how about the enterprising agent that walks through vintage auto show and captures detail about the autos, owners and insurance needs and provides information on the spot.  Or a claims adjuster who immediately responds to auto damage claim and can issue a payment to a customer bank account.

End-of-day processing becomes real-time processing, enhancing customer satisfaction and reducing claims processing costs. Direct and independent agents can provide illustrations, execute underwriting rules, and capture electronic signatures using HP handhelds, tablet PCs, and digital pen and paper technologies.

Now, take it one step further and include predictive modeling on the quote process.  The big test will be teaching the agent how to gracefully tell the prospective account that they can't be written!

Help Wanted: Old Time Claims Guy

Now you can have the gnarly, arms crossed, scowling claim adjuster all your own...with all the knowledge they possess but without the attitude!  Fair Issac has rolled out a product called, MIRA Claims Advisor that captures 10-years of carrier specific data and allows for more accurate setting of reserves.  All the actions of the past, good or bad, will be portrayed in a new reserve model.

It captures all cost drivers, including unique claims characteristics, and considers complex relationships between multiple claim characteristics.

The most important aspect is that it analyzes, using a neural network approach, COMPANY SPECIFIC data.  So far, they found a 20% to 30% in adjuster productivity alone.  A major marketing drawback is convincing carriers claims personnel that aggregate data from bureaus will adequately reflect their exposure base.  Now their history will predict their future, but will they become more comfortable with the mystery of neural networks?

Algorithms on the rise in 2005

Algorithms on the rise in 2005!  Wall Street is still chasing the elusive formula for beating the market.  In 2004,  algorithmic trading accounted for 9 % of all trade...it is "predicted" that in 2005 close to 21% will be done using these techniques ! 

Certainly, the employment outlook for anyone involved in any facet of data mining is on the rise.  But then if the algorithm works, who would need to work?  I guess it is more than buying low and selling hi!

There's Gold in them thar hills!

Gold, not fools gold, but real, honest to goodness riches. As if locked in a vault, insurance carriers, banks, investment firms, or any business for that matter, has a treasure of data to mine in a way to better understand the patterns and behaviors of how and why their customers tick.

As highlighted in Data Mining for Golden Opportunities, there are boundless methods to identify these nuggets.

In the past, mining for gold consisted of choosing a site and then sifting through endless amounts of dirt. Sometimes the prospector only found a few valuable nuggets, sometimes he hit upon an entire vein, but most of the time, he found nothing at all and decided to either move on to another promising spot or give up mining altogether and stop wasting his time. Today, with scientific methods and specialized tools, mineral mining is much more accurate and productive. Mining for data has evolved in much the same way. Older methods executed by business mathematicians and statisticians took a long time to yield constructive information. Now, current software and techniques help make data mining a lucrative, more accessible process for most businesses.

Kudos to the banking industry for attempting to sort through the marketing relationship between savings accounts, insurance purchases, CD and other investments.  Will the insurance industry be able to transition from a focus on underwriting modeling to marketing modeling.  Could there be another gold vein in the mine beyond the underwriting models.  Would strategic relationships between carriers (a PC carrier uniting with a Life carrier to sell each others products) enable its distribution force to be more efficient and focused?  Would profitability come easier with this approach?  After all, isn't it easier to sell an existing customer !

Data Mining Software: Easy-to-Use?

Data Mining Software: Easy-to-Use??  Actually, software available today is designed to be used by the non-experts, i.e., those who do not have an underlying expertise of statistics or mathematics, but DO have both an underlying subject knowledge and basic understanding of data mining concepts.

Data Mining Labs conducted an Evaluation of High-end Data Mining Tools for Fraud Detection looking at the following software vendors: Clementine(ISL), Darwin(TMC), Enterprise Miner(SAS), Intelligent Miner for Data (IBM), Pattern Recognition Workbench (Unica).  The graphic identifies the overall rating for each software application.

The conclusions are as follows:

• IBM's Intelligent Miner for Data has the advantage of being the current market leader, with a strong vendor offering well-regarded consulting support.
• ISL's Clementine excels in support provided and in ease of use (given Unix familiarity) and might allow the most modeling iterations in a tight deadline.
• SAS's Enterprise Miner would especially enhance a statistical environment where users are familiar with SAS and could exploit its macros.
• Thinking Machine's Darwin is best when network bandwidth is at a premium (say, on very large databases).
• And Unica's Pattern Recognition Workbench is a strong choice when it's not obvious what algorithm will be most appropriate, or when analysts are more familiar with spreadsheets than Unix.

Streaming Data

Streaming data allows for near instantaneous decision making for claims management.  In a previous post entitled, "It's Getting Closer...effective mining of Workers' Compensation Claims", it was pointed out that more data was needed to accurately forecast claim severity.  One key element cited was Evidence Based Medicine from the treating physician. 

HM OnLine, a medical bill ASP processor, has developed a solution to this problem.  They have the ability to read 100% of both paper-based bills(HCFA 1500 & UB92) using OCR scanning technologies and of course, digital bills.  But the key to their success is the ability to accurately convert the characters into actual pieces of data that can be assigned to a claim file or through business rules, assigned to a medical case manager, for instance.

Processing patient record information between medical professionals and health care management organization requires timely tenuous phone conversations, photocopying and documentation. The HMonline Electronic Referral and Precertification Management programs establish a portal for streamlining the exchange of information on a real-time basis - 24/7/365. This reduces overhead and enhances efficiency.

The HMonline Electronic Report Submission program allows medical professionals to submit clinical notes associated with the care provided and linked to requests for reimbursement. These documents are connected within the HMonline database and made available to third parties. There is no need for resubmission, additional copying, mailing or distractive phone conversations with third party management organizations. Likewise, management organizations have easy access to vital information without the need for contacting point of service provider offices and can easily communicate this information with legal counsel, case managers, IME vendors and employers.

This streaming data is collected by the physician, at time of treatment, and sent forward to the insurer.  And. the best part...the data is collected and provided for free.  So, perhaps one more piece of the puzzle is solved.

Now all that remains....can you assess the emotional state of an injured worker?