Unlocking the value of health data - Presentation at the Congressional Luncheon Series - Oct 2011
1 like544 views
The document discusses the structure and challenges of managing health data, emphasizing the importance of a data life cycle and layered views of data for clinical and administrative tasks. It highlights ongoing research addressing key issues such as scale, interoperability, security, and sense-making in health data management. The author suggests opportunities to leverage research from other domains to enhance understanding and handling of healthcare data.
Unlocking the value of health data - Presentation at the Congressional Luncheon Series - Oct 2011
- 1. 1 Unlocking the Value of Health Data Sandeep Purao, Ph.D. Research Director, Center for Enterprise Architecture Associate Professor, College of Information Sciences and Technology Penn State University, University Park, PA Congressional Luncheon Series, 5 October 2011
- 2. © Sandeep Purao. spurao@ist.psu.edu My Emphasis is on ‘Data’ 2
- 3. © Sandeep Purao. spurao@ist.psu.edu I have three Goals today • I want to add structure to the large and intractable problem of how to deal with data in healthcare • I would like to point out ongoing research in other domains based on this structure • I would like to suggest how we may leverage this research or identify opportunities for enhancement 3
- 4. © Sandeep Purao. spurao@ist.psu.edu A Data (Life) Cycle helps 4 Generate / Capture Store / Categorize Adding structure to the problem of dealing with data in healthcare Share / Make Available Use / Make Sense Destroy / Shred A data life cycle brings to the foreground the phases through which data must flow
- 5. © Sandeep Purao. spurao@ist.psu.edu A Layered view of Data helps 5 Data for the support of Clinical tasks and systems Adding structure to the problem of dealing with data in healthcare Data for the support of Administrative tasks and systems Data that is Public or is available in Public sources Primary Data from Institutions and Researchers
- 6. © Sandeep Purao. spurao@ist.psu.edu Here is a Simple Structure 6 Generate / Capture Store / Categorize Share / Make Available Use / Make Sense Destroy / Shred Data for the support of Clinical tasks and systems Data for the support of Administrative tasks and systems Data that is Public or is available in Public sources Primary Data from Institutions and Researchers Adding structure to the problem of dealing with data in healthcare
- 7. © Sandeep Purao. spurao@ist.psu.edu Adding Roles to the Structure 7 Adding structure to the problem of dealing with data in healthcare Physicians Patients Policy Makers Lawyers Researchers
- 8. © Sandeep Purao. spurao@ist.psu.edu There are some Key Problems 8 Scale Inter-Operability Using the structure to work with key problems related to data in healthcare Security Sense- Making
- 9. © Sandeep Purao. spurao@ist.psu.edu There is Research Elsewhere • Scale – Big Data – moving Giga to Tera to Peta – Clouds, Hadoop and Map-Reduce – Extracting data from information ** • Inter-operability • Security • Sense-making 9 Pointing to Ongoing Research in Other Domains
- 10. © Sandeep Purao. spurao@ist.psu.edu There is Research Elsewhere • Scale • Inter-operability – Voluntary and Consensus standards including HL7 ** – Heterogeneity, ontology and semantics (NLM) ** – Regional health IT partnerships ** • Security • Sense-making 10 Pointing to Ongoing Research in Other Domains
- 11. © Sandeep Purao. spurao@ist.psu.edu There is Research Elsewhere 11 Pointing to Ongoing Research in Other Domains • Scale • Inter-operability • Security – Dealing with legal co-existence of malicious users ** – Measures such as Role-based Access – Laws to prevent access to EHR ** • Sense-making
- 12. © Sandeep Purao. spurao@ist.psu.edu There is Research Elsewhere 12 Pointing to Ongoing Research in Other Domains • Scale • Inter-operability • Security • Sense-Making – Measuring data quality in crowd-based forums ** – Search patterns and user behaviors ** – Data delivery / use for e-health
- 13. © Sandeep Purao. spurao@ist.psu.edu Opportunities • It is possible to leverage / enhance research from other domains to add to what we know about the Data Puzzle in the Healthcare context 13 Leveraging or Enhancing Existing Research
- 14. © Sandeep Purao. spurao@ist.psu.edu Some Examples - 1 • Example 1: Regional health partnerships – A study of regional health IT partnerships extending ideas and theories about outsourcing – Problem addressed: Data storage and Data sharing • Example 2: Extracting action knowledge – Studies of work in refineries and with health professionals to extract and represent action knowledge – Problem addressed: Data use and sense-making 14 Leveraging or Enhancing Existing Research
- 15. © Sandeep Purao. spurao@ist.psu.edu Some Examples - 2 • Example 3: Changing models for data governance with clouds – A study of healthcare organizations to understand how data stewardship and governance models are changing with cloud – Problem addressed: Data storage and exchange • Example 4: Using context to overcome data heterogeneity – Modeling context to understand how data may be exchanged across different user communities – Problem addressed: Data Sharing and Data use 15 Leveraging or Enhancing Existing Research
- 16. © Sandeep Purao. spurao@ist.psu.edu Some Examples - 3 • Example 5: Data use in collaborative healthcare teams – A study to understand how teams of healthcare professionals access and use data – Problem addressed: Data sharing and Data use • Example 6: Search patterns for specialized information on the web – Empirical analyses of how user communities look for health- related and other information on web sources – Problem addressed: Data Access and Data use 16 Leveraging or Enhancing Existing Research
- 17. © Sandeep Purao. spurao@ist.psu.edu What I said I wanted to do • I want to add structure to the large and intractable problem of how to deal with data in healthcare • I would like to point out ongoing research in other domains based on this structure • I would like to suggest how we may leverage this research or identify opportunities for enhancement 17
- 18. © Sandeep Purao. spurao@ist.psu.edu Summary 18 For further dialog: spurao@ist.psu.edu
Editor's Notes
- #2: I am Sandeep – Sun–Deep Purao – like Perot but spelled differently as you can see. I am on the faculty of the College of Information at Penn State. Would like to begin by thanking Neal for inviting me to a part of this panel of presenters. I am glad to be here and am glad to see a large turnout.
- #3: My mandate today is to talk fairly broadly about the role of data in healthcare. My expertise here is on the word “data” instead of the word “healthcare.” So, with that in mind, I will try to accomplish three goals in the short talk today.
- #4: First – I would like to add some structure to this large and intractable problem of how to deal with data in the context of healthcare. Structure is important because it allows us the ability to identify different components and focus on the one that we deem are critical. Second – I would like to point out some ongoing research in select areas that are suggested by this structure. My intent here would be to share with you opportunities where the healthcare community may be able to leverage some of these research results. Third – I will point out areas where existing research in allied disciplines may not be sufficient or specific for healthcare. This may be one way to distill the arguments in the talk so that efforts and resources may be brought to bear on these areas of research.
- #5: So, the structure I propose is simple. - The first dimension of the structure is the Data Cycle – A data life cycle acknowledges the phase through which data must flow
- #6: The second dimension of the structure are the different categories of data one sees in healthcare settings The first two layers are – clinical and administrative software and adds two others – publicly available data and primary data. These are the horizontal layers. I know this audience knows it well but I will spend a precious minute from the time allocated to me in explicating each. The clinical layer refers to actual patient data such as ailments, medications, vitals etc. The administrative layer refers to data about managing the healthcare delivery system including payments, insurance, schedules etc. The third layer is public data. Consider, for example, webMD, books – increasingly, digitized, physician ratings etc. We can also include publicly available research outcomes such as those in Pubmed or Medline. The final layer is primary data. This is data such as clinical trials (see clinicaltrials.gov), data that researchers collect about obesity and exercise patterns, and data about what makes regional health networks work – collected with surveys and interviews.
- #7: In each cell, I see several opportunities. In each cell, there are also known problems. I would not be here if we did not know the opportunities. I am here because we realize that there are problems that we must overcome to realize the promise of “data in healthcare.” So, let me address some key problems – before I do that I cannot resist the temptation to quickly show you how the simple structure can be seen from different perspectives such as - .
- #8: So a physician may see things differently – as would a patient and a lawyer and a policy maker – These are often implicit – making them open allows us to be clear about where we are focusing
- #9: So here are some key problems then – I have selected four from a long list