Host: Ron M’Sadoques Director of Enterprise Data Intelligence, Hartford Healthcare | As the Director of Enterprise Data Intelligence for Hartford Healthcare, Ron M’Sadoques is responsible for HHC’s Data Management program, including the Master Data Management system, and HHC’s Data Lake and Epic Caboodle Data Warehouse efforts. Ron also is responsible for the ETL functions and procedures.
Ron started his career managing software developers on DEC VAX systems. Over the next 35 years, he managed efforts on CRM and Office Automation systems. Prior to his current role, Ron managed ERP support and development at HHC. |
Session 1: Data Governance Overview
- The difference between Data Management & Data Governance
- Hint: Data Management deals with the structure and performance of the data. Data Governance deals with accessibility and quality of the content of the data
- What is data, and why should it be governed
- An asset that is non-consumable, and retains value
- A PROGRAM, not a PROJECT
- You can stop it when you stop your HR program or Supply Chain program
- Justify it with adverse events
- If data is an asset, then the asset needs to be managed and maintained
- Use the “Truck Fleet Analogy”
- Who are the actors?
- Data Owners, Data Stewards, Application Stewards, Analytic Stewards
- What are the models?
- Depends on the goal.
- If you’re trying to keep your warehouse clean, consider a central model
- If you want better data throughout the org, consider a distributed model
- Depends on the goal.
- Introduce the 6 DG activities
- The difference between Data Management & Data Governance
Session 2: Data Domain Governance
- What is a domain?
- Many disciplines use the term “Domain”. What does it mean to DG?
- Why is Data Domain the first activity?
- The Domain selection process
- Find an easy one
- Find a long-term one
- Find others based upon enterprise strategies and important KPI’s
- Find the Data Owners
- This might be tricky. A Data Owner needs respect and authority across the organization.
- Find the Stewards
- This might be easy. They already exist in your organization, and are recognized in that capacity. Their role is just not formalized
- We’re asking people who spend X hours per week figuring out what happened to spend <X hours per week making sure nothing happens
- What is a domain?
Session 3: Data policy and Strategy
- What is a Data Strategy?
- A set of guidelines and patterns for decision making
- Your Strategy needs to be flexible to changes in business need, and be able to take advantage of new technology
- Data Strategy must follow Analytics Strategy
- Access policies must follow
- Balance the HIPAA “minimum needed” with the organization’s need for insight, and opportunity for monetization
- What is a Data Strategy?
Session 4: Metadata Management
- Easy definition: “The stuff you need to know to make the best use of your data”
- The most esoteric concept to sell to Management
- Critical to democratizing data
- It’s the key to taking the “tribal knowledge” in your Analytics areas, and making it available to all
- Highly participatory
- You will get resistance from people who believe they are the only ones who know enough to access the data.
- The three levels of Metadata
- Dictionary, Catalog, Glossary
- The magic of Data Lineage
- I like the term “data pedigree” – do you really want to make decisions from “mongrel data”
Session 5: Master and Reference Data Management
- The second most esoteric concept
- Initially high investment, subsequent high payoff
- Can greatly shorten time-to-insight
- What is Master Data?
- The 5 P’s – Patients (or People), Providers, Places, Payers, Procedures
- Contrast Master Data in Healthcare versus other industries
- What is Reference Data?
- The obvious – ICD, CPT, SNOMED
- The not-so-obvious – internal codes – encounter types, financial classes,
Session 6: Data Quality Management
- Initially low investment; subsequent high payoff
- Find the adverse events, and springboard off of those
- Data profiling – where it all starts
- How many duplicate keys do you have? How many different gender codes do you have?
- Move to more advanced quality analyses as you know more about the data
- The types of data quality issues
- Validity, Completeness, Accuracy