www.RiskNoodle.com

It's Getting Closer...effective mining of Workers' Compensation Claims

More and more consulting groups and insurance carriers are working on a predictive modeling scenario to identify the severity and length of particular claims.  An October 2002 study by PWC (PriceWaterhouseCoopers) found 83 variables that could be considered potential predictors. Their attempt was to build a model using existing carrier data that would classify claims as "likely" or "not likely" to become serious.  Additionally, data was analyzed using a CART( Decison-tree) model and a logistic regression model.  Their outcomes showed 31% of claims as potentially serious.  However, of that 31%, 2/3rds were false positives. The models also failed to classify 4% of the known serious claims.

So, we still have not found answers to the prediction issues.  Their study points out that additional information such as quantifying the existence of emotional trauma or Evidence Based Medicine from the treating physician could vastly improve the outcomes.

More often than not, injured employees stay out of work longer due to the work environment, the immediate responses of both employer and carrier as well as their internal belief systems.  While each of those variables can be globally controlled, quantifying and capturing these data elements will vastly improve the success of a predictive model.  To further complicate the data collection, how can it be collected in a streaming mode...to allow for immediate triage?

Insurance Agents Work!!

The much maligned insurance agent, believed to be living on a golf course somewhere near you, is actually working.  Agents are the lifeblood of the insurance sales and distribution system.  They own relationships and understand the intricacies of needed coverages.  And, they are also the primary cog in an insurers need to embrace technology.

The sales system is antiquated...relying on old underwriting ideas and using paper (yes, even carbon-paper) submissions that make a tedious task of data transferring, not too mention the error opportunities.  Agent enablement solutions typically combine portal server, content management, and business process management functionality with insurance application modules that can be configured to create agent-broker portals. Key areas of functionality from these technologies include:

• Collaboration tools
• Single view of multiple applications
• Ability to handle transactions
• Value network that enables internal, broker, agent and customer self-service
• Customer-facing functions such as marketing, sales and service via all touch points: in person, by phone, by mail, and over the Internet
• Strong integration with back-end systems
• Ability to develop and adapt business processes, interview, forms and content

Since these agent portals are connected to a carriers backroom functionality, the ability for immediate data mining/predictive modeling present enormous opportunities for "profitable growth".   Using the historical carrier or industry results, a carrier can nearly instantaneously evaluate and price a risk.  As with any data project...success always depends on the quality of the data.  Being gathered and delivered from the primary source can only enhance the sales/distribution system.  Will the extra time from the technology improvement lower the handicap?  Or, will it provide impetus for additional revenues?

This is the voice of Colossus, the voice of Guardian...We are one...this is the voice of unity.

Do you remember the Sci-Fi movie from the late 60's called, Colossus: The Forbin Project (just re-released on DVD)?  It is about a computer that is designed to protect the US from military strikes from the USSR.  Listen to voice of Colossus: it will send chills down your spine.  Colossus is designed with artificial intelligence and quickly merges with the computer from the USSR, called Guardian to become the fountain of knowledge for all mankind...albeit, without freedom. 

So now we hear that Google has planned to index the entire libraries of Harvard, Stanford , The Bodleian and parts of the New York Public Library.

Google's founders, Sergey Brin and Larry Page, have long vowed to make all of the world's information accessible to anyone with a Web browser. The agreements to be announced today will put them a few steps closer to that goal - at least in terms of the English-language portion of the world's information. Mr. Page said yesterday that the project traced to the roots of Google, which he and Mr. Brin founded in 1998 after taking a leave from a graduate computer science program at Stanford where they worked on a "digital libraries" project. "What we first discussed at Stanford is now becoming practical," Mr. Page said.

At Stanford, Google hopes to be able to scan 50,000 pages a day within the month, eventually doubling that rate, according to a person involved in the project.

The Google plan calls for making the library materials available as part of Google's regular Web service, which currently has an estimated eight billion Web pages in its database and tens of millions of users a day. As with the other information on its service, Google will sell advertising to generate revenue from its library material. (In it existing Google Print program, the company shares advertising revenue with the participating book publishers.)

Once this is completed, it may be possible to develop algorithms in predicitve modeling that, by bringing together all the written intelligence of the world, will push science ahead by light years in a matter of hours or days!  Now that is data mining at its best.  But will we live in a Colossus-like world where immediate access to information and its "predictions" will limit our freedoms?

Walmart keeps 460 terabytes of data

460 terabytes of data from every single purchase made at the retail chain.  The data does not focus on the who the customer is, but what "basket" of goods that the customer purchased.  This helps Wal-mart in their approach to aisle layouts and placement of items.  A basket can detail your trip around the store and you don't even know you were followed.  This data is gather by state, region, etc.  Its goal...to manage and tighten its grip on the supply and demand chain... sometimes referred to by Wal-mart suppliers as supply and command.

Dick and Jane…See Spot Run. Run Spot Run !

The simple days of grammar school, now long gone, have given way to a myriad of reforms.  The most recent is No Child Left Behind. 

The legislation is designed to change the culture of America's schools by closing the achievement gap, offering more flexibility, giving parents more options, and teaching students based on what works.

Under the act's accountability provisions, states must describe how they will close the achievement gap and make sure all students, including those who are disadvantaged, achieve academic proficiency. They must produce annual state and school district report cards that inform parents and communities about state and school progress.

States are relying upon data mining techniques to identify districts, school, and classroom performance improvement areas. In Oklahoma, a data warehouse, called Class Server, supports district and state reporting needs and provides analysis capabilities used to drive recommendations for improving student achievement based on numerous data points collected from student systems.

With data warehouse integration, Class Server information is readily correlated with actual measures of student achievement. Oklahoma schools can track the long term effectiveness of changes to curriculum, resources and teaching methodologies for individuals or any subgroup.  Analysis and reporting can be done by administrators using familiar tools and interfaces within the existing Microsoft Office environment. Information for students and parents at home or for the community is easily and securely shared through the Web.

Ebay takes the Pulse of it data

Ebay, the world's largest auction house and 12th largest US retailer, annouced two new services to help its customers track and analyze items that sell on its site.  First, is PULSE, a daily barometer of hot selling items, including vendors and prices.  Can you guess the top selling item???? The I-Pod, surprise!

In addition, a subscription service aimed at sellers was launched by Terapeak, called Ebay Marketplace Services. It provides a more indepth look at sales listing using a metrics based analysis.  It can review millions of transactions to develop and provide up-to-the-minute  ratios, price buckets and length of selling periods.  Think of the analysis used on a stock or portfolio and you will have an idea of the type of data delivered with this service.

Since close to 90% of Ebay sellers are "mom and pop" operations, this information will prove extremely valuable to their success.

Lorem ipsum dolor sit amet

Gibberish, you say.  No! It's Latin.  But, it was actually a Privacy Policy used by a "very large" company.  When you clicked on the "Read the Privacy Policy" button, this Latin (or rather Pig-Latin) placeholder was the only information that was provided.  Seems as though Privacy is the last piece of a web design.

However, there are now Best Practices developed for Privacy Statements. By implementing these practices, companies can gain the confidence of the User, protect the privacy of the data , yet allow for effective data collection to be used for further targeting in underwriting, claims management or marketing.

Here are the Guiding Principles developed by P3P:

"P3P has been designed to promote privacy and trust on the Web by enabling service providers to disclose their information practices, and enabling individuals to make informed decisions about the collection and use of their personal information. P3P user agents work on behalf of individuals to reach agreements with service providers about the collection and use of personal information. Trust is built upon the mutual understanding that each party will respect the agreement reached."

Remember, Garbage In - Garbage Out.  The Critical Success Factors for Predictive Modeling begin with solid, accurate data.  Start with Trust!

Boy, do I need help!

That will be the mantra of the insurance industry for the next few years as they cope with the strategic deployment of IT services for customer interface and internal legacy conversion.   The Customer Respect Center concluded in its Summer 2004 Online Customer Respect Study that over 50% of  74 Fortune 1000 insurance firms studied did not respond at all to web initiated requests.  In addition, there was a significant increase in data sharing between subsidiaries and business partners. (You can review the Insurance Industry report as well as many others.) 

Here is a sampling of their performance on handling your Personal Data:

93% of the companies make mention of how they use personal data

Of these:

• 17% only use it for the purpose it was provided
• 2% do not collect or use data
• 9% use it for internal marketing purposes
• 9% share data with affiliates or subsidiaries
• 24% need to be more explicit about personal data usage
• 36% share the data with business partners, without first getting permission

When compared to other industries, overall the Insurance Group scored a meager 5.7 on a scale of 10. 

In addition, the industry will be relying upon outside help in legacy data migration, backroom development and/or outsourcing, selection of new software platforms, internal training and a much better understanding of using web-based applications for customer interactions.

Handhelds and Healthcare

Handhelds or PDA's come in all shapes and sizes; some with camera, telephones and mp3 players. But the true value with a handheld is to allow instantaneous transmission of data for enhanced decision making. Many healthcare practices are utilizing this workfow technology for recording billable information, accurate HCFA code entries as well as e-Prescriptions. In fact, The Center for Information Technology Leadership, Partners Hospital, Boston, estimates that widespread adoption of e-prescribing could eliminate more than 2 million adverse drug events each year, more than 130,000 of which would be life-threatening. Yet, just over 10% of presciptions filled today, are generated with this approach.

Return on Investment studies on Children's Specialized Hospital of New Jersey indicate an increase in $10 million revenue over the past two years, since each caregiver, equipped with this specialized Palm Pilot can see one more patient per day. In addition, savings have been realized from reduced bill submission from the ususal ten to fifteen to less than 5 days as well as revenue loss from late billing as all data is captured at once.

With this technology used in heathcare for admittedly routine operations, how could a handheld application be used in property/casualty? Could agents be armed with handhelds to canvas neighborhoods to provide on-the-spot home insurance quotations? Part of the answer lies in how you outfit the device. The handheld could identify the property with GPS, capture pre-determined photos of the risk and work "hand-in-hand" with the agent who could visually assess and transcribe the data back to the carriers servers. Once that data is streamed with municipal and credit information, an accurate risk quote could be left for the homeowner... likely within minutes and with all information accurately and precisely documented.

RED SOX ARE WORLD CHAMPIONS!!!

There is joy in Mudville.
Finally, after sitting through 67, 75, 86 and all the other near misses, the Red Sox have won...in four straight.
Now, can they repeat?!? Can data mining be applied...how about predictive modeling in determining hitters for certain situations, or pitchers to face certain batters, but to include various pieces of data, such as weather, temperature, or tracking players behaviors in certain stressful times.

Yes, even baseball has taken its turn at bat in the field of statistics and modeling. Today, they use sabermaticians to calculate the value of a player as measured in wins. Part of that calculation for hitters relies on a calculation of effectiveness in basestealing. Using data from 1962 to 1977, it can be calculated that with a runner on first and no outs, the expected number of runs in an inning is 0.783. That number increases to just over one (1.068) if the runner can steal and end up on second with no outs. With bases loaded and no outs, the expected runs would be 2.254, with one out, 1.546 and with two outs, 0.798.

It should be noted that the Red Sox hired a Sabermatician at the start of the 2003 season, but really, they only needed a LEADER. Thank you Curt Shilling!