Consolidating Information

1
Consolidating Information

• Participants
• H. Burkom - Barbara Spratkes-Wilkins
• J. Coberly - Stella Tsai
• K. Cox
• Questions
• How do we address/resolve the use of multiple
  systems?
• How do we consolidate information, coordinate,
  and communicate across jurisdictions, systems
  analytical results?

2
What is Consolidation?

• Combining diverse sources
• What sources?
• Raw data
• System results (information), ex. alerts
• On same data
• On differing data (same streams), e.g., diagnosis
  vs. CC vs. lab orders vs. drug prescriptions
  within a given system

3
When Do You Consolidate?

• Is it really necessary?
• Need to fuse all results or just representative
  samples?
• At what level is fusion needed?
• Within a system
• Between systems looking at the same data with
  different underlying algorithms
• Between systems looking at different parts of the
  data
• Between systems looking at different regions

4
How Do You Consolidate?

• Optimal Do Not Consolidate
• Consensus Take time, not sure really want this
• 2nd Best (Reality?)
• In your head
• Statistically
• Fusing results from multiple data sources within
  a single system into a single alert
• Some misgivings
• Fusing results from multiple systems,
  meta-analysis-like
• NOT recommended
• Visually with a computer interface showing
  multiple sources/results simultaneously
• Good but have to have drill down capability

5
Rules of Consolidation

• Must lead to actionable information
• Data from the same underlying population
• Must understand the characteristics limitations
  of data being joined
• Data should be complementary
• Carefully evaluate the utility of meta statistics
  
• Must be able to drill down into data details in
  all data sets
• All data sets must (?) offer similar level of
  granularity

6
What Is Needed?

• Common definitions
• Ex. Syndromes
• Have the system in place before an event
• Relationships
• Protocols
• Resources