1.Using Advanced Analytics for Value-based Healthcare DeliveryCopyright Prime Dimensions 2013 All rights reserved.

2. We are a DC-based consulting firm that provides advanced analytics capabilities and implementation services Prime Dimensions offers expertise in the processes, tools and techniques associated with: Data Management Business Intelligence Advanced Analytics We focus on operational aspects and emphasis on Big Data strategy and technology We assist organizations in transforming data into actionable insights to improve performance, make informed decisions and achieve measurable results. We partner closely with clients to ensure cost-effective implementations.Big Data requires a new generation of scalable technologies designed to extract insights from very large volumes of disparate, multi-structured data by enabling high velocity capture, discovery, and analysis. Copyright Prime Dimensions 2013 All rights reserved.1 3. Agenda Challenges & Opportunities Payment Reform Big Data Enabling Technologies Big Data RoadmapCopyright Prime Dimensions 2013 All rights reserved.2 4. Typical Client ChallengesCopyright Prime Dimensions 2013 All rights reserved.3 5. Healthcare in the 21st Century The transformational force that has brought affordability and accessibility to other industries is disruptive innovation. Today's health-care industry screams for disruption. Politicians are consumed with how we can afford health care. But disruption solves the more fundamental question: How do we make health care affordable? Clayton M. Christensen, The Innovator's Prescription: A Disruptive Solution for Health Care National Challenges Escalating Costs Dwindling Budgets More Oversight and Scrutiny Expectation of Transparency and Accountability New Laws, Regulations and Polices Focus on PerformanceTransformational Opportunities Radical innovation Go beyond measuring outcomes to changing them Aggregate and/or discover data in ways that were not possible until recently Visualize problems in humanfriendly formatsCopyright Prime Dimensions 2013 All rights reserved.4 6. Current U.S. Healthcare SituationCopyright Prime Dimensions 2013 All rights reserved.5 7. Medicare Accounts for Over 20% of Total U.S. Healthcare Spending in 2010Note: Other health insurance programs includes DoD and VASource: Medicare Payment Advisory Committee, June 2012 Data BookCopyright Prime Dimensions 2013 All rights reserved.6 8. Healthcare Spending as a Percentage of GDP In 2010, total U.S. healthcare spending was $2.6 trillion, representing 18% of Gross Domestic Product, doubling over the past 30 years. If this trend continues, estimates are that healthcare spending will reach a staggering $5 trillion, or 20% of GDP, by 2021. Moreover, patients outof-pocket costs have doubled over the past 10 years. Source: Medicare Payment Advisory Committee, June 2012 Data BookCopyright Prime Dimensions 2013 All rights reserved.7 9. Transforming the Healthcare EcosystemCopyright Prime Dimensions 2013 All rights reserved.8 10. Agenda Challenges & Opportunities Payment Reform Big Data Enabling Technologies Big Data RoadmapCopyright Prime Dimensions 2013 All rights reserved.9 11. The DNA of Healthcare DataCopyright Prime Dimensions 2013 All rights reserved.10 12. Unsustainable Fee-for-Service (FFS) Payment Model Current FFS payment structure results in redundant testing, medicalerrors, and over-utilization of the system Maximizes providers fees and reimbursements based on volume Incentivizes multiple tests and procedures, regardless of necessity Limited coordination of care across provider network Medical errors and failed procedures result in higher fees for providers Data overload: vital health information assets are not being leveraged 96% of Medicare dollars account for treating chronic illnesses, withonly 3-5% for prevention (Source: CMS, Chronic Conditions Chartbook, 2012)Research indicates that payment reform would reduce spending by $1.33 Trillion by 2023 and significantly improve quality and outcomes. (Source: The Commonwealth Fund Commission on a High Performance Health System, 1/13)Copyright Prime Dimensions 2013 All rights reserved.11 13. Four Major Payment Reform Models Global Payments A single PMPM payment is made for all services delivered to a patient, with payment adjustments based on measured performanc e and patient risk.Bundled PaymentsPatientcentered Medical Home A single A physician bundled practice or payment, w other hich may provider is include eligible to multiple receive providers in additional multiple payments if care medical settings, is home made for criteria are services met, based delivered on quality during an and cost Copyright Prime episode of Dimensions 2013 All rights reserved. performancAccountab le Care Organizati ons Groups of providers that voluntarily assume responsibilit y for the care of a population of patients share payer savings if they meet quality and 12 cost 14. ACO Implementation ProcessCopyright Prime Dimensions 2013 All rights reserved.13 15. Solution We selected three CMS programs for analytics: Accountable CareOrganizations (ACO), Bundled Payments, and Patient Centered Medical Home. The focus of Phase 1 of the prototype is on ACOs.Sources: Bundled Payments for Care Improvement (BPCI) Initiative: General Information, http://innovation.cms.gov/initiatives/bundled-payments/ Against this backdrop, CMS has launched new delivery and payment initiatives aimed to improve patient outcomes and quality while containing costs. We have selected the following programs for this business case: Bundled Payments, Accountable Care Organizations, and Patient Centered Medical Home. Bundled Payments for Care Improvement (BPCI) Initiative: General Information, http://innovation.cms.gov/initiatives/bundled-payments/Copyright Prime Dimensions 2013 All rights reserved.14 16. Payment Reform Solution Helps healthcare payers find the optimal cost, revenue, and providerincentive models to improve health outcomes and contain patient cost Embeds business rules and algorithms of the Medicare Shared Savings Program (MSSP) Accountable Care Organization (ACO). Includes the following features and capabilities: ACO benchmarks and budget based on historical cost baseline, trend estimates and risk adjustments Performance monitoring of key measures and metrics related to cost, utilization and quality Predictive modeling to determine the proper mix of inputs to maximize payment incentives Dynamic dashboards and visualizations to perform trade-off analysis and scenario planningCopyright Prime Dimensions 2013 All rights reserved.15 17. Methodology and Scope The team acquired and loaded public datasets from CMS and related sources Phase 1 scope: In-patient Medicare data for 2010 Sampling of 5 metropolitan areas representing a range of regions and population sizes (CA, FL, MA, OH, GA) Diabetes, hypertension, coronary disease, and heart failure measures Readmissions and ER utilization Created scenarios for comparative analysis and benchmarking Population, cost, clinical and utilization data Analytical models and predictive modeling Seeking input from SMEs Applying industry best practices Assessing technical solutions Data Sources 1. Data Entrepreneurs Synthetic Public Use File 2. Institutional Provider & Beneficiary Summary 3. Data for ACO Applicant Share Calculations 4. CMS Statistics reference booklet for the Medicare and Medicaid health insurance programs 5. National Health Expenditure Accounts (health care goods and services, public health activities, government administration, the net cost of health insurance, and investment related to health care) 6. Premier Hospital Quality Incentive Demonstration (expanding the information available about quality of care and through direct incentives to reward the delivery of superior quality care) 7. Hospital Compare (including 27 quality measures of clinical process of care and outcomes) 8. Assorted files from the CMS Download Database 9. The Dartmouth Atlas of Health CareCopyright Prime Dimensions 2013 All rights reserved.16 18. Agenda Challenges & Opportunities Payment Reform Big Data Enabling Technologies Big Data RoadmapCopyright Prime Dimensions 2013 All rights reserved.17 19. Big Data Enabling Technologies New systems that handle a wide variety of data fromsensor data to web and social media data Big Data requires a new Improved analytical capabilities (sometimes called generation of scalable advanced analytics) including event, predictive and technologies designed to extract meaning from very text analytics large volumes of Operational business intelligence that improves disparate, multi-structured data by enabling high business agility by enabling automated real-time velocity actions and better decision making capture, discovery, and analysis. Faster hardware ranging from faster multi-coreprocessors and large memory spaces, to solid-state Massively Parallel Processing drives and virtual data storage Hadoop and MapReduce Cloud computing including on-demand software-as- NoSQL Databases a-service (SaaS) and platform-as-a-service (PaaS) In-Memory Technology analytical solutions in public and private cloudsCopyright Prime Dimensions 2013 All rights reserved.18 20. Massively Parallel Processing Massively parallel processing (MPP) is the coordinated processing of a program by multiple processors that work on different parts of the program, with each processor using its own operating system and memory Known as loosely coupled or share nothing systems Enables analytic offload Permits parallelized data loading, i.e. E-L-T vs. E-T-LCopyright Prime Dimensions 2013 All rights reserved.19 21. The Hadoop Ecosystem Compute Cluster Multi-structured Source Data Master Node NameNode DataNodeSlave NodeJobTracker TaskTrackerDataNode TaskTrackerSlave Node DataNode TaskTracker Slave NodeMapReduce EngineDataNode Slave NodeTaskTrackerYARNDataNode TaskTracker Slave Node DataNode TaskTrackerDATA LAYERWORKLOAD MANAGEMENT LAYER APPLICATION LAYER Copyright Prime Dimensions 2013 All rights reserved.20 22. The Hadoop Ecosystem Compute Cluster ApplicationMasterMulti-structured Source DataResourceMaster Master Node NameNode DataNodeSlave Node NodeMaster DataNodeJobTracker TaskTrackerTaskTracker Slave Node NodeMaster DataNode TaskTracker Slave Node NodeMaster DataNode Slave Node NodeMaster DataNodeMapReduce EngineTaskTrackerYARN MapReduce 2.0TaskTracker Slave Node NodeMaster DataNode TaskTrackerWORKLOAD MANAGEMENT LAYER Copyright Prime Dimensions 2013 All rights reserved.21 23. NoSQL (Not only SQL) Databases Like HDFS, NoSQL databases have distributed, schema-less data structure and high scale out across low cost commodity hardware NoSQL databases are not reliant on a relational, fixed data model and Structured Query Language (SQL) NoSQL databases are not ACID-compliant and lack the strict data integrity of relational databases NoSQL databases have BASE properties Basically Available, Soft state, Eventual consistency NoSQL databases are queried via REST APIs.Key-ValueColumnarDocument/ObjectGraphIncreasing Data ComplexityCopyright Prime Dimensions 2013 All rights reserved.22 24. In-memory Database An in-memory database stores (IMDS) data entirely in main memory using NVDIMMs, as opposed to disk-based storage Storing and retrieving data in memory is much faster than writing to and reading from disc Reduce data access and latency caused by I/O operations and significantly speeding query and response times.This trend toward in-memory is made possible by low memory prices and multi-core 64-bit processing capabilities. Data stored entirely in-memory is immediately available for analytic processingCopyright Prime Dimensions 2013 All rights reserved.23 25. Database OptionsData VelocityScale UpAnalytic DatabasesRelational DatabasesNoSQL DatabasesScale Out Data Variety Copyright Prime Dimensions 2013 All rights reserved.24 26. Agenda Challenges & Opportunities Payment Reform Solution Prototype Big Data Enabling Technologies Big Data RoadmapCopyright Prime Dimensions 2013 All rights reserved.25 27. Our roadmap reduces risk while progressing towards a unified analytics environment Continuous improvement & innovationChanges to laws (e.g., ACA)Permanent budget pressureInformation Management Health AssessorDesign Thinking & PrototypingMission Oriented Analytics FrameworkUnified Analytics EnvironmentEvolving On-Demand InfrastructureCommodity ITOpen source toolsCopyright Prime Dimensions 2013 All rights reserved.Smarter, mor e demanding data consumers & partners 28. Your Brain on Big Data Query & ReportingDiscovery & EngagementCopyright Prime Dimensions 2013 All rights reserved.27 29. Growth of Data Exploding Eric Schmidt: Every 2 days we create as muchinformation as we did up to 2003. Relational databases can not effectively ingest, store andquery the increased volume, variety and velocity of Big DataCopyright Prime Dimensions 2013 All rights reserved.28 30. Begin with the End in MindCopyright Prime Dimensions 2013 All rights reserved.29 31. Cycles of Design Thinking and PrototypingCopyright Prime Dimensions 2013 All rights reserved.30 32. Mission Oriented Analytics FrameworkCopyright Prime Dimensions 2013 All rights reserved.31 33. Evolving On-Demand Infrastructure for Big Data Advanced AnalyticsAnalytic Offload Scale-up/Scale-outYARNStormNoSQL DatabaseTezAnalytic Applications REST/JSON APIsDashboards & Visualizations Analytic Database E-L-TIn-memory ColumnarHCatalogData Warehouse AugmentationOLAPData WarehouseE-T-LMulti-structured And Stream Source Data Copyright Prime Dimensions 2013 All rights reserved.Data DiscoveryStructured Source Data 34. Unified Analytics EnvironmentCopyright Prime Dimensions 2013 All rights reserved.33 35. Data Management | Business Intelligence | Advanced AnalyticsCopyright Prime Dimensions 2013 All rights reserved.34 36. Questions? Our Contact Information Michael Joseph Managing Partner Direct: 703.861.9897 Email: mjoseph@primedimensions.comRichard Rowan Managing Director Direct: 703.201.2641 Email: rrowan@primedimensions.comPrime Dimensions, LLC www.primedimensions.com Data Management | Business Intelligence | Advanced Analytics Follow us @primedimensionsCopyright Prime Dimensions 2013 All rights reserved.35