A New Approach to Enterprise Data Warehousing for Healthcare Organizations

1
A New Approach to Enterprise Data Warehousing for
Healthcare Organizations

• Steve Eisenberg, MD
• Medical Director, BCBSMN
• David Robinson
• VP Chief Technology Officer, Phoenix Healthcare
  Intelligence

2
Data Warehousing is a Journey
3
Background

• Throughout the history of systems development
• Primary emphasis has always been on operational
  systems and the data they process
• Focus on performance
• Low tolerance for system degradation
• Information needs were an afterthought and always
  came second
• Requirements almost opposite to transactions
• Archived data
• Flexibility
• Broad scope

4
Legacy Systems

• 1970s
• IBM mainframe
• 1980s
• Mini computers
• AS/400
• VAX/VMS
• 1990s
• UNIX platforms Client/server
• 2000s
• Variety of platforms and architecture
• But, over 70 of business data for large
  corporations still resides in the mainframe
  environment
• Speed/Comfort
• Systems built around and house complex series of
  business rules and knowledge
• Difficult to port to new platform

5
Enter the PC

• As business becomes more complex and more
  competitive, the need for information in the
  hands of decision makers grows
• The appearance and subsequent growth in numbers
  and power of desktop computing opened the door
  for business analysts to store, analyze, and work
  with data extracted from legacy systems
• Upside is more information in more hands
• Downside
• Data is fragmented and very source and
  specifics oriented
• Data is not normalized (different answers to same
  question)
• There is no ability for a single data source to
  fit the needs of multiple people
• Analyst must spend a lot of time managing the
  data

6
Early 90s DSS and EIS

• As computer power grew, two strategies appeared
  to meet the information needs for the corporation
• Decision Support Systems
• Targeted towards mid level and lower management
• Very detail oriented
• Executive Information Systems
• Targeted towards senior management
• High level views of data
• Multidimensionality but limited drilldown
• Concepts were excellent but technology
  was not up to the job

7
Early 90s DSS and EIS

• Although they did not succeed, these two
  strategies formed the basis for data warehousing
• Data is preprocessed under a standard set of
  business rules
• Metadata consists of single definitions for all
  data elements and data is named and structured
  for use by non-technical people
• Specific aggregate views of the data are
  available consistent with the needs of the
  organization with the ability to drill down as
  needed

8
Improvements in Technology, Architecture, and the
Plummeting Price for Computer Horsepower (Moores
Law) is What has Allowed Data Warehousing to
Flourish
9
Factors in the Growth of Data Warehousing

• Economic Downturn in late 80s
• Downsizing
• New leadership
• Business process re-engineering
• Consolidation
• All forced corporations to re-look at how data
  was being accessed and reported in the context of
  increased needs with decreased resources

CPU, Disk, Memory Power
Desktop Power Ease
Server Power Ease
Hardware Prices
Software Prices
taken from An Introduction to Data Warehousing
Vivek R. Gupta, Senior Consultant
10
Mid 90s Data Warehousing Begins Coming into
its Own

• Data Warehouses vs Datamarts
• Static Views
• Refreshed infrequently
• Client/Server Architecture
• Desktop access through SQL

11
The Movement to Active Data Warehousing

• Active data warehousing is an evolution of data
  warehousing that moves the concept from simply
  fostering the ability for strategic decision
  making to focusing on both
• The continued development of the data warehouse
  as a strategic business resource
• Coupled with a focus on execution that integrates
  the overall business strategy of the organization
  
• This maturation takes a series of natural steps
  

12
The 5 Stages to Active Data Warehousing

• Stage 1 Reporting
• Stage 2 Analysis
• Stage 3 Prediction
• Stage 4 Operationalize
• Stage 5 - Activate

from The Five Stages of an Active Data
Warehouse Evolution Stephen Brobst and Joe
Rarey, NCR Corporation
13
Stage 1 - Reporting

• Focus is on reporting from a single source of
  truth within the organization
• Major challenge is data integration from multiple
  sources and adoption of metadata and a single set
  of business rules
• Huge Value in bringing disparate sources of info
  from across the organization into a single
  repository
• Drives decision making across functional/product
  boundaries
• Questions tend to be pre-defined
• Huge amount of work but the basis
  for all improvements and evolution
  going forward

14
Stage 2 - Analysis

• Drill down on reports to slice and dice data
• Begin to focus less on what happened
• More on why it happened
• Ad hoc analysis becomes more important
• Because queries are less predictable, performance
  and tuning become more important
• Optimizing the database, efficient and
  sophisticated joins, indexing, etc.

15
Stage 3 - Prediction

• As the what and why become answered, the
  organization will typically want to start looking
  at and understanding what if, or what should
  we expect?
• May involve true predictive modeling using
  sophisticated algorithms but more commonly will
  encompass data modeling
• Workloads accompanying model construction and
  scoring can be huge
• Data mining becomes a term and goal
• Multiple tools available for this
• Though restricted to a few power users, the needs
  of these types of queries can easily overtake the
  capacity of the data warehouse for storage and
  cycle time
• Due to the complexity of the analysis and the
  volume of data

16
The Evolution
Stage 1
Stage 2
Stage 3
Focus on Strategic Decision Making
Stage 4
Stage 5
Focus on Tactical Decision Support
17
Stage 4 - Operationalize

• Stage 4 defines the beginning of active data
  warehousing with the change in focus to tactical
  decision support
• Provides access to information for immediate
  decision making in the field
• Example retail/manufacturing
• just in time delivery of inventory
• Routing and scheduling of deliveries
• Information must be extremely up-to-date
• Refreshing moves to near real time with
  continuous data acquisition
• Query response times need to very fast
• Small number of seconds

18
Stage 5 - Activate

• Movement to tactical decision making raises
  complexity issues and consistency issues
• Natural evolution is to automate those decision
  processes not requiring human decision input
• Decisions become executed with event-driven
  triggers to initiate fully automated decision
  processes
• Example in food industry electronic shelf
  labels where price can be changed manipulated
  from a single central computer
• Pricing changes could be initiated automatically
  within the store based on selling history to
  maximize sell thru and minimize loss of profit
  margins
• An active data warehouse delivers information and
  enables decision support throughout an
  organization rather than being confined to
  strategic decision-making processes
• Supports both tactical and strategic decision
  making

19
The Evolution of the HIG The Rise of (the)
Phoenix
Next
20
Health Information Gateway An Enterprise
Approach to Healthcare Decision Support
6810 New Tampa Highway ? Lakeland, FL 33815 ?
Tel [phone] ? Fax [phone] www.phoeni
xhi.com
21
Our History

• Founded in 1991 in Czech Republic to capitalize
  on opportunities in software development and
  technology
• Built large scale data warehouses for a variety
  of industries
• Data warehouse for a consortium of 13 insurance
  companies processing 100 million transactions
• Created specialty applications for healthcare
  payors
• Individual Risk Scorer Disease Risk Forecaster
  Clinical Pathways Generator Executive Dashboard
• Created a healthcare data model (enterprise data
  warehouse) with a tightly integrated suite of
  applications the Health Information Gateway

Mission To redefine decision support in the
healthcare industry.
22
Typical Current Analytical Environment
How Health Care Executives Currently Obtain
Information to Solve Problems
Day 7
?
Day 5
Day 2
Day 1
Management Question Emerges -Outliers? -Trends?
Get data back and realize you didn't ask the
right questions!
Ask others to research
Search for data
the typical process to turn data into useful
information and knowledge is time consuming,
cumbersome and delay-prone. Many report that
they spend 80 of their time gathering data and
20 analyzing and transforming it into actionable
information.
23
Health Information Gateway (HIG)

• An enterprise-wide platform for decision support
  that allows integration ACROSS business functions
  rather than a point solution

Point-solution vendors e.g., MedStat, Ingenix,
etc.
People in this industry think physician
profiling is decision support. Its not! True
decision support needs to span the enterprise and
to support cross-functional business
needs. --Medical Director, Large Health Plan
24
HIG gives an organization all the flexibility of
multiple datamarts views using an EDW core
model -- access a single version of the truth
25
Browser Based Gateway to Health Information

• Workbench
• Core Business Information
• Point Click Interface
• Just-in-time Dashboard Reports
• Reports from any system

ProAnalyst
Embedded Business Models

• Health Data Analyzer (HDA)
• Total Information
• Broad to specific views
• Knowledge Refinement through Custom Dynamic
  Filters allowing user to apply findings from one
  search to multiple searches

Health Data Analyzer
Claims Analysis
Financial Management
Workbench
Dashboard Reports
Care Management
Utilization Management

• Embedded Business Models
• Workbooks using customer-specific business logic
• Seamlessly apply business rules

Document Management
Provider Analysis
Disease State Mgmt
Provider Profiling
Health Risk Assessment

• ProAnalyst
• Often used in the background of other modules to
  apply business logic
• Communicates directly with HDA
• Stand-alone module with plug-ins for
  sophisticated analysis

Individual Risk Scorer
Disease Risk Forecaster
Clinical Pathway Generator
26
HIG Application Suite powerful set of mining and
analytical tools designed specifically for
healthcare
27
HIG enables users to drill across, down, up and
within multiple business functions
28
HDA - Data flow
HIG Data Flows
29
Technology Behind The HIG

• NCR Teradata
• Performance
• Scalability
• Reliability
• MicroStrategy's Intelligent e-Business
• Generates highly efficient and optimized SQL code
• Powerful Internet tools
• ROLAP

Technology Migration Enterprise Data Warehouse
Event-Based Active Data Warehouse
30
HIG Data Warehouse

• Why NCR Teradata?
• Over 6 years of trying the wrong platforms (3
  Major platforms OS RDBMS)
• While other platforms performed adequately for
  simple models, Teradata excelled on large scale
  in-house benchmarks. Query times on single
  reports for large client went from 24 hours to 27
  minutes
• Unparalleled Query Performance for complex
  Healthcare models
• Scalability, we could grow it (MPP)
• Self Managing

31
Software and Hardware Roadmap
32
Data Transformation and Loading
33
Multi-tier Architecture
34
HIPAA Compliance
35

36
Value Proposition

• Answers at your fingertips with data as current
  as you would like. . . On your desktop . . . Real
  time, any time.
• Platform for easy-to-use, interactive,
  sophisticated analysis delivered through a web
  browser.
• Allows senior executives to tailor their own
  dashboard reports quickly and intuitively.
• Transforms data from all business areas into
  meaningful information, readily available to any
  user, with minimal training.
• Permits you to share information among multiple
  constituents, including providers, driving
  increased efficiency and effectiveness without
  compromising quality.
• Flexibility to travel down any data path with the
  click of a mouse, allowing you to analyze and
  solve complex business problems quickly and
  profitably.
• Users can design ad hoc reports and get results
  within minutes not hours or days.
• Leverages your legacy systems by harnessing the
  power of your current data.
• Implementation in 120 business day cycles to
  realize quick ROI.
• Can be purchased on a license or ASP basis.

The HIG lets executives solve problemsnot just
worry about them.
37
The Power of the HIG Transforming Healthcare
Data into Knowledge
On to the Demo ....