The Future of Data: High-Value Data is the Next Big Thing
1 like585 views
The document discusses the future of data and the need to move from simply collecting data to utilizing high-value data. It notes that the COVID-19 pandemic highlighted issues with timely access to the right data. Key learnings include: improving data acquisition, breaking down data silos, and improving user trust in data. The vision is outlined as moving from static to fast-acquiring data, siloed to integrated data, and untrusted to a single source of truth. Important assumptions driving product directions are also discussed, focusing on healthcare data being a critical asset and analytics converting data into insights.
The Future of Data: High-Value Data is the Next Big Thing
- 1. © 2021 Health Catalyst The Future of Data: High-Value Data is the Next Big Thing TJ Elbert, SVP & GM Data
- 3. Agenda • What have we learned? • Where are we going? • Our vision for data • How will we do it? • Delivering high-value data
- 5. Just In Time Is No Longer Good Enough Access to the right data at the right time is an issue for healthcare. Over the last eighteen months Covid-19 has increased visibility to the issue of high-value data for decision making. • Data acquisition takes too long • A lack of common definitions makes sharing insights a challenge • Integrating data is a challenge outside of a common data model • Poor data quality impacts the readiness of data The data challenges are not limited to the response to Covid-19, and are surfaced in several other use cases • Population health and the shift from fee-for-service to value-based care • Moving from inpatient focused analytics to the inclusion of outpatient insights • Increasing merger and acquisition activity in the industry
- 6. Improve Data Acquisition Static data provides some reporting value, but to unlock high-value data it needs to be readily available . • Changing care models necessitating a variety of data sources • Growing security concerns from granting access across systems • Complex reporting needs bogging down a system • Integrating data through report creation creating redundant work 6 Static data to fast acquisition
- 7. Breaking Down Data Silos Siloed data that has been collocated in a single system provides some insight, but to unlock high-value data it needs to be integrated • Limited view of the patient or member through a single system • A single patient may come across as multiple patients when viewing siloed systems • Layering in labor, revenue cycle, supply chain data, etc. can provide additional value • A lack of a common data model can make report creation a challenge across similar source systems 7 Siloed data to integrated data
- 8. Improve User Trust In Data Data that has been collocated into a single system and integrated can provide a great starting point, but to unlock high-value data you need to ensure the overall quality • Data that is not fit for purpose provides little value • Generated reports, insights, and metrics will get little traction without trust in the underlying data • Transparency into the transformation and quality process is a must to promote trust 8 Untrusted data to a single source of truth
- 10. Poll Question 10
- 12. © Health Catalyst. Confidential and proprietary. Our Starting Point: Success in Healthcare Payment Volume Expansion Labor Supply Chain Population Health Patient Safety Clinical Operations Other You have increasingly critical business issues to address. The most successful organizations focus on revenue, cost, and quality. The biggest wins are integrated and interdependent. Data and analytics are foundational capabilities for these wins. Increase Revenue Decrease Cost Improve Quality
- 13. © Health Catalyst. Confidential and proprietary. Important Assumptions Driving Our Key Product Directions 13 Healthcare Data In the new digital era, data is your most important, mission-critical asset to enable healthcare success Data volume, sources, and uses are growing quickly. • Population health, M&A, affiliates, multi-EMR • Need to improve our data ingestion and processing Successful organizations focus on “high value” data • Integrated data unlocks your highest value use cases • Reusable and transparent data multiplies value • Aggregated data enables benchmarking value • Application data is highly valuable and should be reusable • Data quality is the foundational to “trust” your data • Combining data dimensions produces exponential value Application success will be driven more by data than features. The winning data platform must support all these use cases. • Data volume, high-value data, data-driven applications Healthcare Analytics For critical business issues, analytics is your most important skill set to convert your data into insights Effective healthcare analytics are still too difficult. • Too time and people-intensive • Too SQL focused, not enough analytics focused • Limited available of healthcare domain analytic experts • Difficult to integrate into existing workflows • Difficult to train (costly, slow) • Too easy to misinterpret and get wrong answers • Can’t keep up with data growth • Net: Analytics is difficult to productize and scale AI gives us a unique opportunity to: • Productize analytics & data science capabilities • Get better answers more consistently • Scale advanced analytics across your organization • Answer a broader set of business-critical issues
- 14. Our Product Strategy 14 Market-leading innovation in healthcare data and analytics In Our Data & Analytics Platform • A modern, enterprise-wide platform is foundational to enable data and analytics success • The market is recognizing this with multiple platform models • We believe we have the winning model • Open – supports the broad variety of standard and custom use cases (data, analytics, applications) • Modern, performant, scalable – supports high-growth, high-value data, and analytics needs • Healthcare-specific – supports the complexities of healthcare • Trusted – confidence demonstrated in over 270 case studies In Our Applications • Our applications will address the most important revenue, cost, quality use cases • Our applications will lead by integrating the best data and analytics • Application success will be correlated to the strength of the underlying data and analytics healthcare data and analytics
- 15. • The average hospital has affiliated provides using 16 different EHR vendors • The average health system has affiliated provides using 18 different EHR vendors 15 Sullivan, Tom. “Why EHR data interoperability is such a mess in 3 charts.” healthcareitnews.com, HIMSS, 05/16/2018, https://www.healthcareitnews.com/news/why-ehr-data-interoperability-such-mess-3-charts Data Acquisition
- 16. Reusable Data 16
- 17. Reusable Data 17
- 19. FHIR and Interoperability Standards Expanded DOS Marts provide the foundation for true analytic and transactional interoperability of data via an expanded data model and deep commitment to scalable terminology normalization • First phase supports CPCDS regulations that go into place July 1 • The next phase of DOS FHIR enhancements will add support for USCDI elements − Allergies and Intolerances − Assessment and Plan of Treatment − Care Team Members − Clinical Notes − Goals − Health Concerns − Immunizations − Laboratory − Medications − Patient Demographics − Problems − Procedures − Provenance − Smoking Status − Unique Device Identifiers − Vital Signs U.S. Core Data for Interoperability (USCDI) Common Payer Consumer Data Set (CPCDS) − Patient − Organization − Practitioner − Coverage − Pharmacy − EOB Inpatient − EOB Outpatient − EOB Professional/Non-clinical
- 21. Expanded and Enhanced Data Model 21
- 23. Standard Process: Acquisition • Data acquisition starts at the source and ends when Health Catalyst delivers integrated, reusable, scalable data asset • Health Catalyst has a growing library of 350 data source connectors and data quality checks for its healthcare data model • Health Catalyst stood up a data specific business unit in early 2021 • Consolidated all functions of the data pipeline into a single team • Data acquisition includes a variety of different strategies • Direct database connections • Flat file ingestion • Streaming data ingestion • HIE data ingestion 23
- 24. Standard Process: Standardization • Terminology standardization uses a common code set to create standard reference content • Applies standard sets and attributes (ICD, CPT, MS- DRG, etc.) • Allows for the grouping of codes (Health Catalyst Clinical Improvement Hierarchy, Value Sets) • Makes interoperability, standardization, and governance easier to achieve • Leverages publicly available and standard content classification • Develops data standards that create consistency in the data set • Health Catalyst has developed additional Terminology tooling to aid in value set creation • Automated terminology mapping/Map Manager • Value Set Builder 24
- 25. Standard Process: Integration • The integration of data allows the blending of data from multiple sources of a similar grain and type • Empowers identity resolution (Master Data Management) – Use deterministic matching to merge, cleanse, and standardize data to create a more comprehensive view of a patient or provider • Increases the breadth and depth of insight that can be generated • Allows for more automated measure creation across numerous programs 25
- 26. Standard Process: Data Quality Framework • Data quality is key to fostering a sense of trust in your organization’s data and analytic insights • Missing or incorrect data can: • Remain hidden until your team has headed in the wrong direction • Take weeks or months to track down the source and fix • Consume resources and delay progress • Destroy trust in your organization’s analytic insights 26
- 27. Standard Process: Data Quality Framework HEALTH CATALYST leverages an adaptive library of assessments to validate and monitor data quality; provides expert services Data Knowledge Adaptive Assessment Library: Data quality checks that codify data knowledge, surface hidden issues or quality changes, and avoid repeat issues. Data Quality Deployment Pipeline: Remotely deploy the Adaptive Assessment Library to position checks throughout each client’s data pipeline. Monitoring Active Monitoring: Data quality monitoring results are centralized and enable Health Catalyst teams to validate and monitor data quality and provide robust support. Expert Services Data Quality Services: Our Data Quality team provides consulting and training on best practices; helps define, build, and govern data quality checks; and provides custom services. CLIENTS can build their own data quality program in DOS Review Data Profiles Atlas: Review standard data profiles in the Atlas Data Catalog. Define and Organize Data Quality Checks SAM Designer: Define and organize custom data quality checks that output to a standard data model, allowing results to be surfaced in Atlas and Operations Console. Review Reporting . Atlas and Operations Console: Assess and monitor results of custom and standard data quality checks over time in the Data Quality Assessment worklist.
- 28. Standard Process: Data Quality Framework
- 29. Standard Process: Data Quality Framework
- 30. Value Driven Expert Data Collections • Expert Data Collections • Combination of our expert healthcare data model with a suite of curated data content, such as value sets, populations, and metrics. • Tuned to a variety of healthcare solutions to help you build a sustainable data management system for the future needs of healthcare. 30
- 31. Value Driven Expert Data Collections • Expert Data Collections • Data management strategies to support a rapidly shifting future • Compounded value from integrated data • Solution to challenges of acquiring, integrating, or sharing high quality, timely data • Need to spend less time managing data complexity and get more time to manage data insights 31
- 32. Health Catalyst Expert Data Collections™ Naming Healthcare Foundations Collection Population Health Collections 1. Source Mart ingestion o Clinical o Billing o NPI registry 2. Health Catalyst EMPI 3. Terminology Mapping o Person attributes o Provider attributes o Encounter attributes 4. DOS Marts Model o Person (episode of care elements) o Provider o Person/Provider relationships o Organizations and locations o Encounter o Diagnosis (admit & discharge) o Procedure o Lab results o Medications o Immunizations o Allergies o Charges 5. Core KPI data o Length of stay o Inpatient days o Readmissions o Total charges o Total payment o Volume metrics 6. Data Quality and Performance Optimization • 100+ Data Quality Checks 7. Data and Model Maintenance & Updates Stratify Collection Financials Collection Care Management Collection 1. Value Sets o CMS Chronic Condition Warehouse 2. DOS Marts Model o DOS Risk o Contract Enrollment 3. Claims and Clinical Integration 4. Chronic Condition Populations • Asthma • Alzheimer’s Disease and Related Dementia • Arthritis (Osteoarthritis and Rheumatoid) • Atrial Fibrillation • Autism Spectrum Disorders • Cancer (Breast, Colorectal, Lung, and Prostate) • Chronic Kidney Disease • Chronic Obstructive Pulmonary Disease • Coronary Artery Disease (CAD) • Dementia; Cognitive Decline • Depression • Diabetes • Hepatitis (Chronic Viral B & C) • Heart Failure (CHF) • HIV/AIDS • Hyperlipidemia (High cholesterol) • Hypertension (High blood pressure) • Ischemic Heart Disease • Osteoporosis • Schizophrenia and Other Psychotic Disorders • Stroke • Comorbidity Population: anyone with two or more of the chronic conditions 5. Risk Models o LACE o Charlson-Deyo o Elixhauser 6. Pre-built templates 7. Data and Model Maintenance & Updates 1. Source Mart ingestion o Payer claims 2. DOS Marts Model o Payer claims § Claim header § Claim line § Claim diagnosis § Claim procedure § Member 3. Core KPI data • Member months • PMPM • Readmissions • Inpatient utilization • ED utilization • E/M utilization • High-cost services/imaging • Radiology utilization • Lab/pathology utilization • Post-acute care utilization 4. Benchmark data • Touchstone data • Third-party data 5. Data Quality and Performance Optimization 6. Data and Model Maintenance & updates 1. Source Mart Ingestion • Care Mgt data source(s) 2. DOS Marts Model • Care Managers • Care Team relationships • Care Mgt problems • Care Mgt assessments • Care Mgt goals • Care Mgt programs • Care Mgt interventions 3. Care Mgt KPI data • Patients per care mgr • Duration • Enrollment days • Number of patients enrolled • Attrition rate • Enrollment rate • Dropout rate • Graduation rate 4. Data and Model Maintenance & updates Regulatory Quality Collection Ambulatory Quality Collection 1. ECQMs • 22 certified (subset) 2. HEDIS Measures • 40 certified (subset) 3. MIPS Measures • 100+ (subset) 4. Data Marts Models • DOS Measures • Contract Enrollment 5. Data and Model Maintenance & Update 6. Tailored data services • TBD 8. Tailored data services (applies to all) o Core data model extensions o Additional data models o Additional sources o Custom metrics o 3rd party data integration o 3rd party application integration (i.e. ACG grouper) o Real-time data o External updates
- 33. Value Driven Expert Data Collections • Expert Data Collections Benefits • Reusable content foundation across a diverse set of data • Improve the ability to acquire, integrate, and share high-value data • Provides an optimized data model • Manage data as a strategic asset • Significantly improve time to deliver insights 33
- 34. Value Driven Expert Data Collections • What is new with the DOS Mart healthcare data model • Improve time to value – Data products that focus on expert data collections – Data quality framework integrated into the acquisition process • Support for regulatory quality reporting – Expanded ambulatory content – Augmented intelligence for terminology mapping • Increased data integration and scale – Parallel loading, expanded content – DOS Marts on the Snowflake data cloud • Interoperability Standards – FHIR (Cures Act) 34
- 35. High-Value Data • High-value data requires timely acquisition, standard applied definitions, a flexible model to integrate, and wrapped in a robust data quality program • To bring high-value data to Health Catalyst clients, we are bringing innovation into our approach, organization, and data model • New tooling, an updated engagement strategy, and targeted data acquisition strategies will move our clients from data to high-value data 35