State of Texas

ADVANCE PLANNING DOCUMENT (APD)

HHSC Office of the Inspector General (OIG)
Fraud, Waste, and Abuse Reduction Initiative:
Graph Pattern Analysis and Fraud Case Management Software

Submitted to:

Centers for Medicare and Medicaid Services, Region VI
August 2012

Submitted by:
Texas Health and Human Services Commission

 Table of Contents

SECTION 1 - INTRODUCTION ...............................................................................................................................3
1.1
1.2

PURPOSE OF THE ADVANCE PLANNING DOCUMENT UPDATE ...........................................................................3
BACKGROUND OF TEXAS MMIS ......................................................................................................................3

SECTION 2 - NEEDS AND OBJECTIVES ..............................................................................................................5
2.1
2.2
2.3
2.4

BACKGROUND ..................................................................................................................................................5
BUSINESS NEEDS..............................................................................................................................................6
SOLUTIONS PROCUREMENT PROCESS ...............................................................................................................7
COST BENEFIT ANALYSIS (CBA) / BENEFITS OF THE PROJECT.........................................................................8

SECTION 3 - DESCRIPTION OF ACTIVITIES, PROPOSED SCHEDULE, AND STAFFING ......................8
3.1 OVERALL SCOPE ..............................................................................................................................................8
3.1.1
AUTOMATED FRAUD DETECTION ................................................................................................................8
3.2
TIMELINES ................................................................................................................................................. 13
3.2.3
TOTAL TIMELINE FOR COMPLETION – ALL PARTS..................................................................................... 15
3.3 STAFFING ....................................................................................................................................................... 16
SECTION 4 - BUDGET ............................................................................................................................................ 16
4.1 BUDGET ......................................................................................................................................................... 16
4.1.1
LYNXEON – GRAPH PATTERN ANALYSIS SOFTWARE ............................................................................... 16
4.1.2
IBM CASE MANAGEMENT SOFTWARE, DATA WAREHOUSE AND IDENTITY INSIGHTS ............................... 16
4.2 FEDERAL FINANCIAL PARTICIPATION ............................................................................................................ 16
SECTION 5 – ASSURANCES .................................................................................................................................. 16
5.1 CFR ASSURANCES ......................................................................................................................................... 16
5.2 COMPLIANCE WITH CMS SEVEN (7) CONDITIONS AND STANDARDS .............................................................. 16
5.2.1
MODULARITY STANDARD ......................................................................................................................... 17
5.2.2
MITA CONDITION ..................................................................................................................................... 17
5.2.3
INDUSTRY STANDARD CONDITION ............................................................................................................ 19
5.2.4
LEVERAGE CONDITION .............................................................................................................................. 19
5.2.5
BUSINESS RESULTS CONDITION ................................................................................................................ 19
5.2.6
REPORTING CONDITION ............................................................................................................................. 20
5.2.7
INTEROPERABILITY CONDITION................................................................................................................. 20

 Section 1 - Introduction
1.1

Purpose of the Advance Planning Document Update

The Texas Health and Human Services Commission (HHSC) submits this Advance Planning
Document (APD) to request federal financial participation (FFP) from the Centers for Medicare
and Medicaid Services (CMS) to implement a comprehensive fraud case management software
toolset in support of its Fraud, Waste and Abuse reduction initiative.
This APD requests funding to modernize the entire OIG case processing system, using graph
pattern analysis as the cornerstone of an investigative process based nearly exclusively on
analytics. This investigative process complements the traditional reactive approach to
complaint-based investigations with an active approach that identifies patterns of fraud and abuse
before they become apparent.
Although OIG will still accept and process investigations
originating with complaints, OIG intends for the backbone of its case identification method to be
graph pattern analysis.
HHSC submits this APD-U in accordance with 42 CFR Subpart C 433.112, 42 CFR Subpart C
433.117 and 45 CFR 95.605. These Federal regulations allow State agencies to request FFP for
costs incurred for the operation of the MMIS to improve the efficiency and effectiveness of the
service delivery of the Medicaid program. HHSC submits this APD to request prior approval to
implement these changes. FFP for work at the enhanced 75/25 and 90/10 match rates are
requested for this project.

1.2

Background of Texas MMIS

The Texas Medicaid program has operated since September 1967 under Title XIX of the Social
Security Act, as amended. The program is funded by appropriations authorized by the Texas
State Legislature for each biennium. The current biennium is September 1, 2011 through August
31, 2013. In the current biennium, state/federal funds for Medicaid are projected to comprise 32
percent of all-funds state expenditures (about $172.3 billion).
HHSC is the single state agency responsible for the State of Texas Medicaid program and has
final authority for Medicaid policies and changes made to the Medicaid Management
Information System (MMIS). The MMIS is a distributed group of procedures and computer
processing operations and subsystems. HHSC contracts with a coalition of vendors, headed by
Affiliated Computer Systems State Health Care, LLC (ACS) and working under the name of the
Texas Medicaid & Healthcare Partnership (TMHP) to provide MMIS services. The contract
between HHSC and TMHP authorizes TMHP to provide the necessary services to process and
adjudicate Medicaid claims (with the exception of capitated arrangements between health plans
and HHSC and pharmacy claims). Compass21 is the name of the MMIS system used to support
the processing and adjudicating of claims.
Texas Medicaid serves a total population of approximately 3.6 million eligible recipients per
year and an average of 2.7 million in any given month. TMHP performs services to support the
following claim and non-claim related areas of MMIS operations:

 










Compass21
o
Clearinghouse (Compass21 Electronic Data Interchange (EDI) Gateway) –
Provides Health Insurance Portability and Accountability Act (HIPAA)-compliant
/ EDI transaction routing
o
Recipient Subsystem – Supports recipient and eligibility management
o
Provider Subsystem – Supports provider enrollment and information management
o
Reference File Subsystem – Supports tables’ maintenance for edits and audits
o
Provider Portal – Web-based interface for Providers to access information and
perform eligibility verification and claims’ status inquiry
o
Core Claims and Encounter Processing Subsystem – Supports editing, auditing
and pricing of claims and encounters
o
Third Party Recovery System (TPR) – Support third party recovery and liability
processing
o
Cash / Financial Subsystem – Support cash and financial processing
o
Vision21 (V21) AdHoc Query Platform (AHQP) – Datamart for information
management and analysis of data related to Medicaid claims and associated data
o
Management and Administrative Reports Subsystem (MARS) – Supports defined
MARs and Statistical reporting
o
Surveillance and Utilization Review Subsystem (SURS) – Supports SUR analysis
and reporting
o
V21/Institutional Reimbursement (IR) - Supports specific reporting and analysis
for IR-related extracts and reports, such as Blue Ribbon
o
Automated Inquiry System (AIS) – Supports phone inquiry of claims and
eligibility related information for providers
o
Online interface (Phoenix) – Supports end-user functions for interaction with
claims and related information
Case Management / Health Education – Supports case management and health education
functions for Primary Care Case Management (PCCM) members
Vendor Drug Program (VDP) - The Vendor Drug system includes two subsystems: the
Claims Payment System and the Pharmaceutical Rebate Information Management
System.
Provider Network Management – Supports provider network management, credentialing
and enrollment for PCCM providers
Member Management – Supports member processing for PCCM members
Claims Submission (TexMedConnect) – Provider application that supports claims
submission, eligibility verification and claims’ status inquiry
Claims Management System and Service Authorization System – Long Term Care
service authorization and claims processing engine
Encounters Datamart – Stores managed care encounter data from contracted Managed
Care Organizations.
Vendor Drug system, which currently includes two subsystems: the Electronic Claims
Management System, and the Pharmaceutical Rebate Information Management System.

In addition to the MMIS systems operated by TMHP, HHSC operates the Medical Fraud and
Abuse Detection System (MFADS)

 Section 2 - Needs and Objectives
2.1

Background

The greatest challenge to identifying and interdicting Medicaid fraud is discovery of the fraud
itself. Fraud happens very quickly yet discovery can be slow, resulting in huge opportunities for
criminals and huge costs for taxpayers. In FY 2010, Texas identified over $71 million dollars in
overpayments to health care providers as a result of fraud and abuse. In FY 2011, that figure
had dropped to a mere $28 million. Although identification of fraud and abuse in FY 2012 is
significantly greater than in FY 2011, these overpayments – at any amount – clearly represent the
tip of the iceberg when it comes to the costs of fraud and abuse in Medicaid. HHSC’s Fraud and
Abuse initiative seeks to find the fraud beneath the surface that remains undetected and then
actively manage its anti-fraud process through a fraud case management system.
Currently, the systems and processes within HHSC OIG are built around a complaint-based
investigative process. The addition of active analytic tools to identify fraud as it happens
represents a significant paradigm shift towards a more aggressive approach to recovering
taxpayer money lost to Medicaid fraud and abuse. This approach will complement and reduce
the reliance on the “pay and chase” and complaint-based investigative model of fraud
management.
OIG has determined that a more appropriate response to the ever-changing schemes and artifices
to defraud requires an adaptable, multi-pronged, overlapping approach that provides internal
validation of discoveries while also permitting a rapid response to identify patterns, behaviors or
schemes.
In support of this aggressive approach, HHSC conducted significant research in determining the
best solutions to address undetected fraud, including but not limited to the following:






Research on solutions with the Gartner group;
Discussions with other states on their approaches to F&A identification;
Research of federal agencies, including the Department of Defense and intelligence
agencies;
Internal discussions with Stakeholders and HHSC Leadership; and
Extensive discussions with and participation in proofs of concept with commercial
vendors.

HHSC identified two key needs in the active and aggressive identification, investigation and
management of previously undetected and unreported fraud:



The analysis of existing Medicaid claims data to identify trends through Automated
Fraud Detection;
Active management of identified fraud cases not developed around the traditional
reactive complaint process.

More detail on these needs follows in section 2.2 “Business Needs” below

 2.2

Business Needs

As discussed above, HHSC has identified two key business needs in support of its Fraud and
Abuse reduction initiative.

2.2.1 Automated Fraud Detection
OIG has determined its capabilities are severely limited in detecting and prosecuting the
investigation of fraud and abuse as it is happening. However, using graph pattern analysis we
can see trends develop that we would not otherwise know to watch. Graph pattern analysis is a
highly advanced form of analyzing data that graphically depicts the relationships between and
among virtually any data point: entities, times, places, amounts, services, similarities, absence of
similarities, differences – in short, anything that might represent a relationship of interest. For
example, in Texas, orthodontia spending grew astronomically over approximately five years, an
aberration that went completely unaddressed – even though the fact of increased spending was
both self-evident and discussed. The reason in substantial part was that the Legislature had
increased funding for the Medicaid dental program so everyone, from legislators to program
managers to those charged with ensuring program integrity, expected spending to increase. In
short, the State was able to determine that spending increased but was unable to perceive how it
was increasing: the providers, locations and types of billing were warnings to state officials but
there was no one who could see the greater picture and address the problem. Said another way,
because spending increased uniformly across the orthodontic profession, there were no outliers
to draw attention to an aberration. A rising tide lifts all boats and in this case the entire program
was the aberration. Among other possibilities, graph pattern analysis would have identified the
absence of outliers as a warning, would have identified the spending trend as an aberration
compared to the funding provided in the General Appropriations Act and would have compared
rates of growth in orthodontics to other areas, flagging the profession for further investigation.
Graph pattern analysis permits the user to compile and analyze huge quantities of data – from
Medicaid encounter and billing data, to property ownership and Secretary of State records, to
Dun & Bradstreet and commercially available databases, to Craig’s list and EBay – in
seconds. In the analysis of that data lies the relationships that exist among and between entities,
times, places, actions – or any other type of analysis the user wishes to examine. The primary
advantage graph pattern analysis has over the existing targeted query analysis used in
investigations is that the graph pattern analysis is undirected. It directs the data not toward
targeted queries (“How many times does “X” billing code or event occur in conjunction with
“Y” billing code or event) but toward unspecified analysis (“Show me what providers are doing
similar things and the relationships that exist between them”). It is through this undirected
analysis that HHSC OIG will greatly improve its investigation and fraud recovery ability. Using
orthodontics as an example, graph pattern analysis could have identified that orthodontists in the
Medicaid program were purchasing more objects – property, cars, planes and other items – at a
rate disproportionate to their peers, other providers or indeed, themselves in previous years. In
another example, OIG could identify connections between high frequency events, actors and
locations (suggesting the presence of kickbacks or other incentives). In other words, where
targeted queries ask data to reveal known unknowns, graph pattern analysis asks data to reveal
unknown unknowns.

 2.2.2 Case Management and Reporting
OIG does not have a fraud and abuse case management and reporting system to complement the
automated fraud detection capabilities. In reality, the case management system OIG currently
has does not permit OIG to query its caseload with any degree of automation. For example, it is
impossible to generate an answer to the question, “how many cases did OIG open in the second
quarter of FY 2012 and at what dollar amounts?” without manually opening every case and
checking to see the date the case began and the dollars at risk. Similarly, there is no efficient
way to monitor cases by individual investigator performance, case progress, case size, dollars at
risk, amount of time a case has taken, or any other useful metric. Without a database and
reporting system for fraud case information, OIG is unable to develop performance metrics to
ensure quality fraud case management. Currently, the only useful metrics information OIG can
obtain is via manually culling through case files or by opening each case individually in the
existing case management system, identifying relevant data and tallying the results manually.
A case management system not only will allow parsing of active and historical caseloads to
develop reports and analyze performance metrics. It will also permit investigators to upload
reports, photographs and digital audio/visual recordings from the field, will incorporate report
templates (drastically enhancing the quality of reports and decreasing the time necessary to
prepare them) and will require time management and logging (to determine the amount of human
and financial resources a particular case has required and provide a cost breakdown for
recoupment purposes). Most significantly, a case management system will tie into graph pattern
analytics so that OIG’s own reports will become a database: we will know how many times Dr.
“X” appears in any OIG report (regardless of the context: witness, expert or target), will know
where cases are trending and the types of cases, where MCO’s notice over- or underutilization,
and so on. The proposed case management system will enable OIG to monitor case progress
electronically, meaning managers will receive electronic notifications every time a case has been
accessed, every time a deadline is missed, every time a report is uploaded. It will enable OIG to
process cases more quickly and more efficiently, ensuring OIG investigators meet the agency
target of case completion within eight weeks. Equally important, it will enable investigators to
access their cases from the field, enabling OIG to transition to a more mobile, responsive
approach.
Perhaps most significantly, OIG’s proposed case management system will integrate with two
highly sophisticated analytics programs. In that manner the OIG case data itself will become a
valuable data set, revealing trends (across geographic areas, population centers, medical
specialties, provider type, billing code – any data point captured in an investigation) and making
possible a rapid response to identified suspicious behavior that previously would not be
identified at all, except by chance.

2.3

Solutions Procurement Process

After more than a year of evaluating various proposals and technologies, HHSC OIG determined
that no single approach can entirely capture the scope and breadth of waste, fraud and abuse in
the Medicaid system.
Based on identified needs, HHSC developed a Statement of Work it used to initiate a search via
Texas’ existing collective purchasing vehicle managed by the State Department of Information
Resources (DIR). HHSC OIG reviewed several vendors via a competitive, best value

 procurement which provided the Statement of Work to vendors with offerings within specific
standard industry classifications.
After a thorough review of responses, HHSC OIG determined the above needs as described in
Section 2.2 could be satisfied by two vendors:



Automated Fraud Detection- 21st Century Technologies using LYNXeon Graph Pattern Analysis
Software
Fraud Case Management, Reporting and Analysis- IBM Investigative Case Management
Solutions

Proposals for each of these two vendors are attached to this APD, and a scope summary of these
solutions are described below in Section 3.1 “Overall Scope.”

2.4

Cost Benefit Analysis (CBA) / Benefits of the Project

HHSC OIG has determined several key benefits to a joint implementation of the above solutions
from 21st Century/LYNXeon and IBM.








An aggressive, active approach to detecting fraud;
The ability to identify and open additional cases through graph pattern analysis;
Better information to close cases and support successful convictions;
Prevention of dollars lost to fraud and abuse by recognizing the loss far sooner than current
processes allow;
Relief from the complaint based, “pay and chase” investigative process;
The ability to report on performance metrics for fraud cases; and
Enhancing efficiency by identifying cases with high probabilities of fraud, waste or abuse prior to
expending time and financial resources.

Section 3 - Description of Activities, Proposed Schedule, and Staffing
3.1

Overall Scope

The following represents a summary of scope of the two solutions represented in OIG’s Antifraud and Abuse initiative. Specific scope surrounding each of the solutions can be found in the
attached proposals for 21st Century (21CT)/ LYNXeon and IBM respectively.

3.1.1 Automated Fraud Detection:
21CT/LYNXeon Graph Pattern Analysis
The central component of the analytic solution for HHSC is 21CT’s LYNXeon Automated Fraud
Detection software, a high-performance, enterprise-class platform that provides powerful and
customizable graph pattern fraud analytics, combined with flexible and intuitive
visualization. LYNXeon offers the benefits below.

 Automated Fraud Analytics and Workflows
LYNXeon provides a powerful system based on Graph Pattern Analysis and analytical
components offered through the LYNXeon Analyst Studio. Unique analytic features for fraud
detection include the Analytic Catalog, Fraud Cluster Analysis, Fraud Association Discovery,
and Advanced Analytics using Pattern Query Language. These capabilities allow the fraud
investigator to query the data in a way that reveals patterns and relationships between people,
places, events, times, and things. They also allow queries based on groups, “nearness” and other
clustered or networked behavior. These varied analytic techniques offer investigators choices in
how to uncover connections in seemingly unrelated data.
The fraud investigator starts the investigation in the LYNXeon Analytic Catalog. The process
begins with selecting one of many pre-identified LYNXeon fraud analytic approaches or by
selecting an Analytic from results that the system has already run and return. As LYNXeon
users find useful patterns of suspicious treatment, improper billing, outlier behavior or other
remarkable situations, they can then develop their own approaches, or add further layers of
queries to existing queries in order to identify other clues to suspicious activity. There is always
the opportunity to share new queries with other analysts and schedule queries to run
automatically. The Analytic Catalog provides an organization a means to collaborate and diffuse
expertise between both expert and novice fraud investigators. The result is an organization with
supporting technology that can leverage the human intellect of tens or hundreds of analysts,
building upon experience and each query to develop thousands of automatic queries completed
in seconds, as often and regularly as desired.
Cluster Analysis provides a groundbreaking, patented clustering technique that detects
significant relationships without pre-existing knowledge of categories or classifications. In
contrast to conventional statistics-based clustering techniques, LYNXeon’s Cluster Analysis
inspects the link-based connectivity of data elements to reveal groups of items that appear related
to one another, such as groups of providers or patients who may be colluding to defraud the
agency. More simply, rather than relying upon statistics and algorithms to identify when things
look aberrant (compared to established norms), Cluster Analysis mimics human thought by
looking at the relationships between and among groups of elements. Knowing that ten providers
are the highest billers of a particular suspect code is interesting, but easy. What is particularly
helpful is to know what led to them being the highest billers: who do they know? How and when
and with whom do they bill? What is connected to their billing? Based on these links, what
other connections can we expect? Cluster Analysis affords this type of intelligence.
Meanwhile, Association Discovery explores correlated HHSC data and discovers more complex
chains of connections between data points such as people, places, times and things. So for
example, consider a physician whose patients consistently use the same pharmacy to fill
prescriptions. This type of activity could be indicative of fraud between the physician and the
pharmacy yet is difficult to find with directed data queries. There are links in the data from the
physician to the patients and the patients to the pharmacy, but no direct links to tie the physician
to the pharmacy. Association Discovery can uncover these patterns of activity by linking
together intermediary data elements and revealing those patients who use the same physician and
get their prescriptions filled at the same pharmacy and location. Manual and human
interpretation of these patterns of activity could take analysts months; Association Discovery
reduces this time to seconds, even on massive data repositories.
Advanced Analytics allows for the extension of the LYNXeon platform to serve domain-specific
needs. Thus, common and repetitive tasks or workflows useful to the HHSC analysts can be

 simplified to one-step mouse-click actions in LYNXeon’s user interface, Analyst Studio. This
not only expedites the process for investigators (thus saving time for more important activities)
but it also serves as a reminder to investigators to do basic investigative work on a regular basis
(and makes it easy to do so).
LYNXeon Analyst Studio is a workstation tool for accessing all of the components provided by
the LYNXeon Enterprise Server platform. Using Link Explorer—a flexible and intuitive
canvas—analysts can visually interact with data in ways never before possible. HHSC analysts
can discover and visualize relationships and links in the data, query and ask questions of the data
without having to understand complex languages or algorithms, observe the progression of
events over time, and overlay data on maps using the integrated GIS viewer. Aside from being
one of the most powerful tools on the market (and the only capable of graph pattern analysis)
LYNXeon is also intuitively easy. One of LYNXeon’s strengths is its practical application:
nearly any investigator or analyst can learn to use it easily.
3.1.2 Solution Integration
1: Automated Fraud Detection
and Fraud Investigations
Performed in LYNXeon
Analyst Studio

2: LYNXeon integration with
IBM Case Management for
building reports

LYNXeon Analyst Studio

LYNXeon Enterprise Server

Case
Reports

IBM Case Management System (ICM)

LYNXeon
Graph
Data
Warehouse (IBM provided
High-performance
Data
warehouse)

Other
HHSC
Provided
Data Sources Provided
Data Sources

3: ICM and
ICM
and Identity
FAMS Insight
data
augment augment
data
and enhance and
the
data LYNXeon
enhance
the data
analyzes
LYNXeon
analyzes

IBM InfoSphere
Identity Insight

Figure 1: Proposed Architecture and Usage Diagram

In order to understand the OIG solution, it is important to recognize the OIG proposal draws on
the strengths of two apparently divergent (and otherwise competing) technologies. The
LYNXeon approach relies upon a mathematical graph of information upon which the tool and
users can perform analysis to detect fraudulent behaviors. IBM will provide two critical
systems: Identity Insights, which provides entity resolution (clarifying and cleaning data to
eliminate errors attributable to corruptions or repetitions in the data itself), and IBM’s Case

 Management System for case reporting. Both systems will also serve as data sources for
LYNXeon analysis (see Figure 1).
In the example of a Google word search, the traditional approach yields a query like this:
“identify all the Medicaid fraud experts in Texas,” where the LYNXeon approach yields a query
like this: “identify all experts in the available data.” The first query would result in a list of
names answering a specifically defined search. Depending on the data available, the second
might result in a graph of every expert contained in the data, identifying the linkages between
those who work in Medicaid cases and those who specialize in other areas, defining those who
testify and those who do not, showing linkages between higher paid experts and their clients,
permitting the user to see relationships between and among experts, clients, dates, nature of
testimony, and so on.
The OIG proposal contemplates a broad spectrum approach, utilizing and integrating ordinarily
competitive business products into a single, comprehensive analytic approach to interdicting
waste, fraud and abuse in Medicaid expenditures. Once fully integrated and implemented, OIG’s
proposal will draw on the strengths of divergent approaches and technologies, fusing them into a
powerful, comprehensive method of organizing, storing, accessing and easily manipulating and
utilizing enormous quantities of data. The result will be an aggressive approach to early
identification and prevention of Medicaid overpayments. Nothing like this approach exits in
federal or state Medicaid policing. The closest similarity to the capability described in this
proposal, to analyze structured and unstructured data with both directed and undirected queries,
can be found in major credit card companies and the National Security Agency.
In brief, OIG proposes to marry technologies – initially, by positioning LYNXeon to utilize
IBM’s data warehousing capacity as OIG’s primary data storage solution. Requiring LYNXeon
to take advantage of the IBM data storage technology ensures the two systems are linked from
the outset of the implementation. Then, LYNXeon will integrate IBM’s Identity Insights and
IBM’s Case Management System (CMS) products as source data, actually integrating and
ingesting them into LYNXeon’s Automated Fraud Detection system.
IBM’s InfoSphere Identity Insights
This effort will involve pulling data from a variety of data sources, some provided by HHSC and
some provided from other external sources. LYNXeon’s graph data warehouse is essentially a
correlation and fusion of these data sources into a linked graph of this source data. This graph
representation is necessary to conduct automated fraud detection using graph pattern analysis.
However, as is often the case, the same real-world entity (a person, for example) can be
represented as multiple aliases across several data sources (“Dr. X” may appear that way in
Medicaid data sources, as “Mr. X” in academic or Secretary of State records, but as “John and
Mary X” in county property records. A solution must exist to resolve and disambiguate data
entities.
IBM InfoSphere Identity Insight is a real-time, scalable entity resolution and analysis platform.
Its pioneering identity and relationship disambiguation technology combined with innovative,
complex event processing will ensure the best data source for analysis in LYNXeon as shown
below in “ICM and Identity Insight data augment and enhance the data LYNXeon analyzes” of
Figure 1.

 Advanced Case Management
An advanced case management system is integrated into the OIG proposal though IBM’s Case
Manager (ICM). This platform provides an environment where agencies and programs can
customize solutions to support specific mission areas, uniting information, processes and people.
The ICM solution affords OIG a comprehensive view of case information to achieve increased
performance through performance metrics, time and expense tracking, case identification
methods (that is, recognizing which cases fit queries such as “how many cases have potential
exposure greater than $5 million and were opened after January 2012?”) and other access points
to cases and case data that OIG determines will optimize outcomes. ICM allows OIG
investigators and managers to extract and manage critical case information through integrated
business rules, collaboration and analytics. These capabilities enhance decision making ability
and lead to more successful case outcomes.
ICM is a highly flexible and adaptable solution that provides a foundation for capturing and
replicating the best of the organization’s practices. Through templates, the ability to autopopulate those forms with data available from LYNXeon-utilized databases and an extensible
infrastructure for meeting specific program and departmental priorities, ICM provides a complete
approach to case management: OIG will be able to manage – and most significantly, access –
cases and the data they contain through a number of queries limited only by the needs of the
agency. ICM and its case analysis elements are designed to be reused within parallel or future
solutions, thereby further reducing risk, cutting costs and speeding the time-to-value for the
software. Said another way, ICM can integrate with the same databases LYNXeon uses, autopopulate forms with routine data (e.g., identification numbers, addresses, corporate ownership)
and thereby save hours of investigator time. Later in the process, investigators can use ICM
loaded templates to prepare case reports, load audio visual recordings and access case files from
remote locations.
The IBM proposed solution includes migration of cases from OIG’s existing case management
system, IFM, in two stages. In this way, OIG will be able to maximize the value of ICM by 1)
maintaining historical case management data, and 2) enhancing the value of that data by being
able to access it readily, query it and use it in LYNXeon analytic approaches. OIG will integrate
LYNXeon with IBM’s case management system by both ingesting data from the system as well
as feeding LYNXeon analytic results back into the system, creating a symbiotic relationship
between the two technologies. This symbiosis will have the effect of creating additional
valuable data points for LYNXeon to analyze (permitting investigators to see patterns and trends
in OIG’s own cases among case types, actors involved, geographic centers of activity and
various other measurements) while simultaneously providing investigators the ability to use ICM
to provide access to cases in a method that empowers OIG investigators to extract and manage
critical case information through integrated business rules, collaboration and analytics. Further,
ICM will offer OIG previously unavailable access to historical case information from the
existing case management system. OIG will be able to leverage that historical case information
immediately and enhance the value of that data by being able to access it readily, query it and
use it in the LYNXeon analytic approach.
OIG assesses IBM’s Case Manager as the most effective, customizable and practically useful
case management application available, easily integrated with LYNXeon in a manner to add
value to both technologies.

 High-performance Data Warehousing System
IBM will provide a high-performance data warehousing solution to serve as the data storage
medium for LYNXeon Enterprise Server. This data warehouse will hold LYNXeon’s graph
representations of Medicaid claims data, various other database resources (such as Dun &
Bradstreet, Texas Secretary of State data, commercially available databases, property records and
other sources of information) and, in time, information originating from SNAP/WIC/TANF data
as well as cyber network data. This data warehousing solution has the capacity to accommodate
large scale data analysis (tens or hundreds of terabytes) and the expected user workload of even
an OIG office enhanced with additional investigators and system users.
Assumptions
HHSC will provide access to the data sources used by both 21CT and IBM systems. OIG will
provide the fraud investigators, physicians and nurse expertise.

3.2

Timelines

The integration will take place over a two-year period, as diagrammed in this timeline and
detailed below.

Figure 2: Timeline of Phases and Tasks

 Description of Tasks
Phase 1: Initial Delivery of Automated Fraud Detection and Case Management System
(Months 1-4)
21CT will conduct business analysis to gain an understanding of the processes and workflows
around HHSC fraud investigations. This analysis will drive the design of LYNXeon graph data
schemas and specific fraud analytics for HHSC. Upon completion of the design phase, 21CT
will deploy LYNXeon at HHSC and will ingest Medicaid data into LYNXeon server. 21CT
expects to reach this milestone within the first two months after contract start.
21CT will work with HHSC analysts in months 3-4 for build an initial set of LYNXeon
Analytics and workflows. This will involve training of HHSC analysts and working with those
analysts on an ongoing basis throughout the effort. At the end of this Phase 1 period (month 4),
21CT will deliver repeatable workflows for fraud investigations and HHSC analysts will be able
to demonstrate these capabilities.
Parallel to the 21CT efforts, IBM will implement the base investigative case management
system. This phase will permit the capture and storage of both scanned paper and electronic case
documents (e.g., Word, PDF, PPT) for creation and population of cases. OIG Program Integrity
team members with appropriate access controls will be able to edit and update new versions of
these documents, annotate, comment and modify cases and case documents. IBM will migrate
pending cases from the current case management system to the new case management system
and will enable data reporting for inactive cases pulled from the IFM system. OIG will be able
to create summary reports, statistics an analysis across both current and historic case data. After
observing OIG processes during the initial four months, IBM will identify initial OIG
requirements and will implement IBM Case Manager to provide basic case management
functionality, including roles, content types, forms, workflows and case storage in the IBM Case
Manager repository. This system will go live in month four.
Throughout the Phase 1 of this effort 21CT will conduct requirements gathering for the
integration of LYNXeon with IBM’s case management systems and document an integration
plan. At the end of month four, 21CT and IBM will be able to demonstrate the use LYNXeon
and ICM components necessary to meet Phase 1 deliverables.
Estimated completion: 4 Months after Contract
Phase 2: Integration of LYNXeon and IBM’s Identity Insights and data warehousing
system (Months 5-8)
21CT and IBM will complete the integration of LYNXeon and ICM. Specifically, this will
entail feeding ICM reports back into LYNXeon so that they can serve as data sources for
Automated Fraud Detection and Investigations (see Figure 1: Proposed Architecture and Usage
Diagram). LYNXeon will identify 3 additional data sets as identified and agreed upon by OIG
and 21CT during the Phase I effort (e.g., Dun and Bradstreet data, Texas Secretary of State data,
commercially available databases, property records and other sources of information). IBM’s
Identity Insights will be deployed to perform entity resolution on this data and feed the resolved
data directly to LYNXeon’s graph data warehouse (see Step 3 in Figure 1). Identity Insights will
be delivered at the end of month 5.

 By the end of this Phase (month 8), LYNXeon will fully utilize the IBM provided Highperformance Data Warehouse system. Integration efforts of ICM and Identity Insights data into
the LYNXeon graph data warehouse will also be complete.

Estimated completion: 8 Months after Contract
Phase 3 Analytics Development, System Enhancements, Support and Ongoing Training
(Months 9-24)
If not already finalized, IBM will complete migration of data related to inactive cases from the
legacy IFM case management system, enabling HHSC to decommission the system. This
migration will be complete no later than month nine.
LYNXeon Analytics development will continue and build upon the Analytics work of the first
three phases. This Analytic development work will focus on further automation of the detection
of fraud activities and will be an ongoing task through remainder of the effort.
At the beginning of this Phase, 21CT and OIG will begin a process of development cycles in
which both parties will explore the capabilities of the complete implemented solution, document
capability gaps and needed improvements in an “Enhancement Plan,” and then implement new
capabilities. An example enhancement development cycle would be:


Identify and Document Desired Capabilities



Create/Modify Enhancement Plan



Design



Implement Enhancement Plan



Test



Demonstrate and Train on New Capabilities

At the end of the development cycle, all parties will repeat the process again, identifying new
capabilities, workflows and customizations, based on the latest iteration of the entire system.
Each of these development cycles is expected to last two months with a tight customer feedback
loop at the end of the cycle to drive the next cycle of development.
Estimated Completion: 24 Months after Contract
3.2.3 Total Timeline for Completion – All Parts
The total timeline for completion of this project, all parts is two (2) years from the approval date
of this Advance Planning Document (APD-U).

 3.3

Staffing

Both proposals are turn-key solutions and include all staffing components as part of the overall
purchase price. HHSC OIG anticipates it can accomplish implementation of both of these
projects by reallocating existing State staff.

Section 4 - Budget
SECTION 4.1 REMOVED – PROPRIETARY PRICING INFORMATION
4.2

Federal Financial Participation

Federal Funding Participation Allocation: APDU Related - MMIS
Allocation Category
State Portion
Federal Portion
Total Federal & State
Non-MMIS 50/50 FFP
$
$
$
MMIS Related 50/50 FFP
$
$
$
MMIS Related 75/25 FFP
$
4,080,061.00
$
408,006.10
$ 3,672,054.90
MMIS Related 90/10 FFP
$
15,837,512.00
$
1,583,751.20
$ 14,253,760.80
Total Amendment/COR Cost
$
1,991,757.30
$ 17,925,815.70
$
19,917,573.00

Section 5 – Assurances
5.1

CFR Assurances

The following assurances are provided in accordance with the requirements of 45 CFR 95.613,
and 95.615, and 42 CFR 433.112(b) (5)-(9).
The Texas Health and Human Services Commission agrees that:






5.2

The State shall own any software, procedures, or publications that are designed,
developed, installed, or improved with 90 percent FFP. The State shall retain the right to
sign, extend, and cancel any licenses for software used in operation of the enhanced
Medicaid system.
HHSC has a royalty-free, non-exclusive, and irrevocable license to reproduce, publish, or
otherwise use and authorize others to use software, modifications to software, and
documentation that is designed, developed, installed, or improved with 90 percent FFP.
The costs of the system will be determined in accordance with the principles of fair and
open competition in accordance with 45 CFR 95.613.
All information in the system will be safeguarded in accordance with 42 CFR 431,
Subpart F, or with such standards as established by the Secretary of Health and Human
Services under the provisions of the Health Insurance Portability and Accountability Act
of 1996.

Compliance with CMS Seven (7) Conditions and Standards

The following Seven Conditions and Standards as prescribed in Medicaid IT Supplement (MITS11-01-v1.0) are discussed below:

 5.2.1 Modularity Standard





SDLC and Development. The 21CT/LYNXeon and IBM Case Manager are both
complete Commercial-off-the-Shelf (COTS) products. The products purchased are turnkey solutions built on open platforms such as Linux, and SQL. The State is not
conducting any additional, in-house software development for this solution.
Interfaces: The joint 21CT/IBM solution is completely based around modular, industry
standard API’s which are described in depth within each of the proposals.
Business Rules. The joint 21CT/IBM solution employs a number of open, accessible, and
human-readable business rules to automate the discovery and management of unreported
fraud cases. These business rules can be exported to any HHS designated national
repository.

5.2.2 MITA Condition
In August 2012, Texas Medicaid expects to complete its second MITA State Self Assessment
(SS-A) based on recently released version 3 of the CMS MITA Framework. During the
assessment, the State set specific goals for improvement based on the scale set forth in the MITA
Maturity Model (MMM). To achieve the goals, the State contracted with Cognosante (formerly
FOX systems) to develop three deliverables as part of the MITA SS-A: 1) An “As-is” analysis,
2) A “To-be” roadmap, and 3) A “Gap” analysis which determines the gaps between the “As-is”
state and the “To-be” level of business capability, based on the MMM.
The final submission
of Texas MITA 3.0 SS-A is expected in October 2012.

For this APD, we have listed the relevant MITA business processes that will be impacted by the
benefits of the Managed Care Expansion project in consideration of these recommendations:
MITA 3.0
PROCESSES

BUSINESS

PROBLEM ADDRESSED

ASSISTS
MITA
LEVELS

IN

3.0

IMPROVING
MATURITY

OIG currently has no way to
develop metrics and measure
The 21CT/LYNXeon solution
performance on the identification
raises the MMM rating for
of fraud and abuse in the Texas
this process from Level 1 to
PL05
–
Manage Medicaid system, due to a manual
Level 2 for automation, data
Performance Measures
process that requires extensive
accuracy, effectiveness,
involvement from investigators.
efficiency, and Level 3 for
Performance reporting requires
utility to stakeholders
investigators to manually open
every case to retrieve data.
As stated previously in this
The 21CT/LYNXeon solution
document, from an identification
raises the MMM rating for
perspective referrals are based on
this process from Level 1 to
PE01
–
Identify provider complaints, and there are
Level 2 for automation, data
Utilization Anomalies
issues with respect to this
accuracy, effectiveness,
business process. OIG believes
efficiency, and Level 3 for
that fraud complaints/referrals
utility to stakeholders
represent the “tip of the iceberg”

 when determining the extent of
fraud and abuse.
The OIG
currently has no automated
process to identify patterns of
abuse as they happen.
Medicaid Provider Integrity’s
(MPI) cases are predominately
identified
and
driven
by
complaints. These complaints are
related to providers enrolled in the
Texas Medicaid and/or CHIP
The 21CT/LYNXeon solution
program.
raises the MMM rating for
this process from Level 1 to
Self-generated referrals can be
Level 2 for automation, data
PE02
–
Establish initiated by any investigator,
accuracy, effectiveness,
Compliance Incident
Manager or nurse in MPI who
efficiency, and Level 3 for
comes into information that leads
utility to stakeholders.
us to believe a provider may be
committing fraud, waste or abuse
in the Texas Medicaid Program,
however, the current workload is
too high to review these cases and
supporting information is almost
never readily available.
Compliance
incidents
are
generated using a completely
manual process that is exclusively
reliant on human intervention.
Regardless of the source of the
complaint the Medicaid Program
Integrity Intake Unit and/or MPI
Field staff performs the following
actions:

PE03
–
Compliance
Information

Review the information received.
Research is then performed to
Manage
verify that the complaint is in
Incident
regard to an enrolled Medicaid
provider. Manual verification is
required and is not automatically
cross referenced to a provider’s
information. Once verification is
performed, a manual request must
be submitted for a case to be
opened on the legacy case
management system, which then
generates a case number.
The process is extremely time

The 21CT/LYNXeon solution
raises the MMM rating for
this process from Level 1 to
Level 2 for automation, data
accuracy, effectiveness,
efficiency, and Level 3 for
utility to stakeholders.

 consuming, as it is a completely
manual process reliant on human
intervention in the management
and monitoring of fraud cases.
As stated previously throughout
this APD, incident determination
is a completely reactive process,
utilizing a complaint-based, “pay
and chase” process.
The
determination of an adverse action
incident is a completely manual
process
reliant
on
human The 21CT/LYNXeon solution
intervention. The length of time
raises the MMM rating for
for an investigation has been years this process from Level 1 to
PE04
–
Determine and an efficient tracking, for
Level 2 for automation, data
Adverse Action Incident
referrals and outcomes, system is
accuracy, effectiveness,
required.
efficiency, and Level 3 for
utility to stakeholders.

5.2.3 Industry Standard Condition
Both the 21CT/LYNXeon solution and the IBM Case Management solution are end-user data
repositories that will natively accept X12 claims and encounters information.
5.2.4 Leverage Condition
In accordance with receipt of 90/10 funding, HHSC agrees to share project artifacts with other
states/federal entities to the extent available by law.
5.2.5 Business Results Condition
The combined 21CT/LYNXeon and IBM Case Management solution generates significant
business results through the following:


Automation: The combined 21CT/IBM solution exponentially increases the level of
automation regarding fraud discovery in the OIG’s office through Graph Pattern Analysis
to identify trends and fraud patterns that remain unreported. This information is fed into
the IBM Case Management solution which develops cases and automates workflow for
resolution of these cases.

 

Performance Standards: The IBM Case Management solution will enable automated
tracking, and reporting of case statistics assist in the development of performance metrics
that will be used to measure case performance and fraud recovery activities. Reporting
against these metrics will be available to key stakeholders including Texas HHS
executive leadership, the Texas Legislature, and the Centers for Medicare and Medicaid
Services (CMS).



Customer Service: The combined 21CT/IBM solution best serves our most important
customer – taxpayers. The reduction of fraud and abuse and the subsequent recoveries
will lower the cost of Medicaid for taxpayers, provide opportunities for expanded service
at legislative discretion and enhance access to care by increasing the attractiveness of the
program through enhanced program integrity.

5.2.6 Reporting Condition
The joint 21CT/IBM solution as described completely embodies the intent of the reporting
condition. The 21CT/IBM solution is a scalable solution that uses standard SQL databases as the
foundation of query reporting. Reporting data will be available electronically (see attached
proposals for greater detail)
5.2.7 Interoperability Condition
Since the 21CT/IBM solutions are leading COTS products in the healthcare marketplace, and use
open source technology, as well as open interfaces, we expect interoperability with any national
exchange to be seamless.