Insurance Fraud Detection and Prevention with Sybase IQ solution ...

Insurance Fraud Detection and
Prevention with Sybase ® IQ
Reduce financial cost of insurance fraud, increase speed of investigation,
and improve accuracy of fraud detection
solution overview
www.sybase.com

SOLUTION FEATURES &
BENEFITS AT A GLANCE
• Execute complex queries
across massive data sets,
increasing the accuracy of
statistical models.
• Mine a wide range of
document types for critical
fraud indicators.
• High-speed query processing
ensures answers when they
are needed, enabling fraud
to be detected earlier in the
claims process.
• Data compression keeps large
datasets manageable and
affordable, enabling insurance
organizations to perform
analysis on full and complete
data sets — no need for
samples or pre-aggregation
of data.
Fraud is a well-known problem in the insurance industry. It is costly for both policy holders and insurance
companies in all sectors of the industry. Recent research indicates that fraudulent claims account for a significant
portion of all claims received by insurers, and cost billions of dollars annually. Fraud has been seen to occur at
many different stages in the process of an insurance transaction. It can occur in the application process, when an
individual is a policyholder, as well as by third-party participants. Professionals that provide services to claimants also
sometimes perpetrate fraudulent claims. Some common frauds include misrepresenting information on the insurance
application, inflating actual claims, staging false accidents, and claiming serious injury when only minor injury has
actually occurred.
Additionally, insurance companies feel a sense of urgency to improve their ability to detect and prevent fraud
as the sophistication and volume of fraudulent claims continues to increase. Also, new regulatory requirements,
the negative impact on profitability, and the need to manage a widening range of risks mandate that insurance
companies begin to use more cutting-edge technologies and analyses to battle fraud.
To successfully address these challenges, insurance companies are equipping their fraud examiners with advanced
analytics technologies. Equipped with tools to analyze the organization’s business data, fraud examiners can more
accurately identify transactions and activities that indicate a risk of fraudulent activity earlier in the process. Data
analysis also provides an effective way to be more proactive in the fight against fraud by seeking out indicators
or patterns of fraud in the data. These analyses require the skills of highly trained professionals who are expert at
developing and manipulating complex statistical/mathematical models with data mining and other analysis tools.
However, they also require a high performance analytics server to run these models in real-time against massive data
sets with large dimensionalities.
Insurance companies need an analytics server that gives them the ability to see complex changes, ask difficult
questions, and get to answers faster than any other approach. Additionally, they need to be able to join together
huge quantities of information and ask previously unanswerable questions. This requires the ability to integrate
information silos and use data “as is” to perform cross-sectional analyses. Specifically, insurance companies expect
their analytical models and the server they run on help them achieve four key fraud detection and prevention goals.
To deliver on these demands, insurance providers need:
• Reduce the financial cost of fraud by building the most advanced detection systems that can leverage
all available information.
• Increase the speed of investigation to identify fraudulent claims before payment occurs.
• Improve the accuracy of fraud detection, by reducing both the number of false positive and false
negative detections.
• Minimize operational costs by centralizing data into one data warehouse without resorting to expensive
hardware solutions.
1

When it comes to choosing an analytics server to help them succeed with their fraud detection and
prevention initiatives, insurance companies worldwide have selected Sybase IQ. As a high performance
analytics server designed specifically for mission-critical business intelligence, Sybase IQ enables insurance
organizations to analyze massive data sets in real-time using advanced data mining and statistical techniques,
facilitating more accurate detection and prevention of fraud.
Because of its unique column-based data architecture, Sybase IQ can execute complex queries 50-100x
faster than traditional database systems, and can do so across entire data sets (not just data samples), thereby
increasing accuracy. It can leverage advanced data mining techniques and statistical methods including
supervised learning, neural networks, principal component analysis, and clustering methods, and can also
mine a wide range of document types for critical fraud indicators. Additionally, Sybase IQ can store all data in
one data warehouse without ballooning storage costs, because of advanced data compression technology.
Leading financial firms and insurance companies such as Prudential, Allianz, Citigroup, and MasterCard have
chosen Sybase IQ over other alternatives because of four primary differentiating benefits:
• Performance – Sybase IQ has been optimized for analytic workloads, and provides much higher
performance for solving fraud detection searches than conventional solutions. Sybase IQ’s columnbased
structure, patented indexing technology, and unique query optimizer allow it to deliver superior
query performance, especially on complex or ad hoc queries. Query optimization is achieved through
the use of metadata and multiple column indices in priority order to achieve the fastest response. This
means that more potential fraud cases can be identified sooner and more lengthy histories can be
searched for fraud patterns.
• Accuracy – Sybase IQ speeds design, scoring, tuning, and implementation of predictive models up to
400 times faster on large data sets and complex queries. Sybase IQ supports complex fraud detection
models running against full data sets — no need for samples or pre-aggregation of data which limit
the amount of data analyzed, which results in incomplete or inaccurate analyses. The in-database
analytics capabilities of Sybase IQ are especially important when technical analysis needs to happen
on large datasets and requires high accuracy. Data does not need to be moved between Sybase IQ and
the fraud detection application, so the improvement in query performance is significant and complex
relationships can be identified more accurately.
• Data Types – Sybase IQ can analyze all types of structured and unstructured data types, including
text data. The ability to analyze textual data such as email and other documents is extremely valuable
for fraud detection and prevention. Insurance claims processes collect and generate large volumes
of text-based information, and text analytics software accesses this, parses it to distill meaningful
insights, and analyzes the newly created data to gain a deeper understanding of the claim.
• Cost Effectiveness – Conventional solutions solve analytics problems by using more computer
hardware, often driving up costs for only marginal improvements. Sybase IQ takes a different
approach — it stores the data in a column-oriented architecture and provides sophisticated data
compression which allow it to search data while still compressed. These techniques allow Sybase IQ
to search data sets that are often 50% to 75% smaller than the original raw data, with 10-100x fewer
input/output operations when compared to traditional databases. Data compression saves on storage
and maintenance costs.
“For each individual analytical
query, the operational OLTP
database often took an extremely
long time to return the results. We
wanted to enable our divisional
users to run queries by themselves
and customize those queries as
necessary. To accomplish this, we
needed a system that is easy to
understand, flexible and, above all,
very fast.”
– Andreas Seibert, IT Manager, AOK Hessen
www.sybase.com/detailid=1059440
“We originally projected that
Sybase IQ would save us about
$150K per year, but over the past
five years, it has saved us about
$5 million by helping us identify
forms of fraud we never knew
existed and by giving us the ability
to implement appropriate measures
to eliminate the causes.”
– John Hagan, Project Manager,
American Airlines
www.sybase.com/detailid=210272
2

Health Insurance (Payers)
Using Sybase IQ for fraud analysis, health insurance payers can extract relevant data from claims systems, external
data sources, and unstructured text, as well as import business rules for known fraud, waste, and abuse schemes from
existing rules engines. They can employ advanced scoring based on statistical and predictive modeling techniques
that enable discovery of emerging suspicious claims activity that would otherwise go undetected. Typical statistical
techniques used with Sybase IQ for these purposes include:
• Profiling models that model the behavior of groups or individuals, building models of usual and customary
behavior from history, either for that individual or for peer groups.
• Clustering that identifies abnormal groups of claims, either because they are outliers in every respect, abnormal in
relation to a selected base (such as customer segment or profile), or contain values that are abnormal in relation
to each other.
• Data-mining tools and build programs that produce fraud propensity scores. Adjusters simply enter data, and
claims are automatically scored for their likelihood to be fraudulent and made available for review.
Sybase IQ has been effectively used to detect fraud in health insurance firms for typical usage scenarios involving
member group characteristics analysis, outpatient, inpatient, and integrated care data, analyzing invoices from
doctors, pharmacists, hospitals, and therapists, identifying out of ordinary costs, and performing trend analysis of
scope of services relative to treatment diagnosis.
Additionally, Sybase IQ is part of an integrated ecosystem of technologies that support fraud analysis for health
insurance payers, including:
• Sybase Industry Warehouse Studio for Health Insurers – A pre-built business intelligence application suite that
contains sample data, schema, scripts, and pre-built reports that are based on lessons learned from “hands-on”
experience gained over many years in the health insurance industry.
• SAP® BusinessObjects EIM Data Services – A versatile data movement technology suite that includes a
Data Integrator ETL tool optimized for loading Sybase IQ from a variety of data sources and a robust data
quality module.
• SAP BusinessObjects Enterprise for Reporting – Enables managing performance and governing compliance using
market-leading reporting, charting, and dashboarding tools with highly secure access rights.
3

Property and Casualty
In the Property and Casualty insurance industry, fraud detection is key to improving customer retention
and satisfaction. By eliminating fraud, insurance premiums can be lowered as the portion of the premiums
associated with covering fraud costs is reduced. Also, companies can minimize negative customer perceptions
that can be caused by too much investigation of “false positives” (suspicions of a fraudulent claim that turn
out to be incorrect).
For P&C insurers, devising effective fraud detection systems requires a number of analytical challenges
to be addressed, such as discovering small subsets of claims with a high percentage of recoverable fraud,
isolating the factors that indicate a claim or payment request has a high probability of fraud, developing
rules and using them to flag only those claims or requests that are most likely to be fraudulent, and ensuring
adjusters can review claims or requests that are not only likely to be fraudulent but also have the greatest
adjustment potential.
P&C fraud detection solutions using Sybase IQ can focus on numerous advanced and sophisticated
techniques to drill down on fraudulent activities including:
• Understanding and Capitalizing on Data – Previously audited claims data hold the key to recouping
money in the future. By creating models from historical information, P&C insurers can accurately pinpoint
fraudulent claims out of the millions of claims each year. Also, an important step is to continually ensure
the right data is used to solve the problem. The data should not disproportionately represent any segment
of the population sample being studied for fraud. To effectively test for fraud, all relevant transactions
must be tested across all applicable business systems and applications.
• Building Mathematical Models to Find Fraud – P&C insurers need to accurately determine which claims
are fraudulently filed so they can concentrate on preventing unnecessary claims payment. Over the
years, they collect audit results on insurance claims. Data mining now enables insurers to predict which
insurance claims are likely to be fraudulent. This gives adjusters the power to determine what returns
they should target, thereby, recouping millions of dollars otherwise lost and saving adjusters many hours
of valuable time. These models examine each line entry on claims, compare the line entries against the
amount of fraud dollars detected, and rank claims in the order of likely fraud and display the results back.
“Active data analysis has
increased the amount of
collected customs taxes,
exposed more illegal foreign
exchange, and enabled us to
crack down on smuggling
several times over compared to
the levels achieved prior to the
implementation of the CDW. And
as business investigations are
conducted on the basis of solid
data, our business activities are
streamlined as well.”
– Sang-Chul Park, Deputy Director, Korea
Customs Service
www.sybase.com/detailid=1045450
• Capitalizing on the Value Hidden in Textual Information – Some examples of this capability include
using text mining to look for scripted comments in fender-bender crash claims. For example, identifying
unrelated claimants using the same comments on their claims. Text mining can be very helpful in
revealing these types of discrepancies or conditions.
4

Additionally, Sybase IQ is part of an integrated ecosystem of technologies that support fraud analysis for Property
and Casualty insurance companies, including:
• Sybase Industry Warehouse Studio for Property and Casualty Insurers – This module is an excellent choice for
those insurance organizations that need to implement a working analytics system in the shortest amount of time
with a minimum of risk. The IWS approach to the implementation of a data warehouse for property and casualty
fraud detection allows insurers to take advantage of the opportunity to buy a data warehouse as an effective
alternative to the more traditional approach of designing and building a warehouse on their own.
• SAP BusinessObjects EIM Data Services – A versatile data movement technology suite that includes a
Data Integrator ETL tool optimized for loading Sybase IQ from a variety of data sources and a robust data
quality module.
• SAP BusinessObjects Enterprise for Reporting – Enables managing performance and governing compliance using
market-leading reporting, charting, and dashboarding tools with highly secure access rights.
5

IN SUMMARY
Sybase IQ successfully supports fraud detection and prevention worldwide for insurance companies, and
helps them increase the speed of investigation, improve accuracy of detection, and reduce financial losses due
to fraud. In addition to all of the capabilities enumerated in this brochure, Sybase IQ also offers the following
options which are highly beneficial for preventing fraud:
• VLDB Option: The Very Large Database (VLDB) option enhances the manageability of very large datasets,
which is typical with enterprise-wide fraud management initiatives. Large data sets are critical in looking
for historical fraud patterns in the data.
• Unstructured Data Analytics Option: Textual data such as email and other documents are often analyzed
for fraudulent claims. The Sybase IQ Unstructured Data Analytics option provides an application friendly
and scalable text analysis functionality to detect fraud patterns in unstructured data.
• In-database Analytics with Quantitative Libraries: Performing deep data mining and statistical analysis
is a key aspect of fraud detection. The Sybase IQ In-database Analytics option with the sophisticated
quantitative library — DBLytix from Fuzzy Logix — provides the ability to sift through vast amounts of
data — structured and unstructured — to look for fraudulent patterns.
• Advanced Security Option: The need to provide secure data in-flight and at-rest necessitates use of
advanced capabilities such as strong encryption and authentication. The Advanced Security option helps
prevent unauthorized access to data and reports that opens up opportunities for fraud.
“Most businesses don’t want IT
to disappear into a back room
for nine to 12 months, then come
out and say ‘Look what we’ve
done! Businesses want visibly
fast results, which is what we’ve
given them with Sybase IQ and
Industry Warehouse Studio. From
October to December we achieved
the complete implementation
of our data warehouse
infrastructure, complete with
scalability to support our future
needs. This was probably one
of the fastest data warehouse
implementations in history.”
– Scott Wyld, Project Manager, Allianz
Australia
www.sybase.com/detailid=1033709
6

To learn more about how
Sybase IQ supports fraud
detection and prevention
please contact your Sybase
sales representative or call
1-800-8-SYBASE.
Sybase, Inc.
Worldwide Headquarters
One Sybase Drive
Dublin, CA 94568-7902
U.S.A
1 800 8 sybase
www.sybase.com
Copyright © 2011 Sybase, Inc. All rights reserved. Unpublished rights reserved under U.S. copyright laws. Sybase, the Sybase
logo and Industry Warehouse Studio are trademarks of Sybase, Inc. or its subsidiaries. ® indicates registration in the
United States of America. SAP and the SAP logo are the trademarks or registered trademarks of SAP AG in Germany and in
several other countries. All other trademarks are the property of their respective owners. 09/11
BusinessObjects is a trademark of BusinessObjects Software Ltd. BusinessObjects is an SAP company.