CTMS Metadata Efforts

1
CTMS Metadata Efforts

• City of Hope National Medical Center
• Division of Information Sciences

2

• Metadata Project
• Generating Common Data Elements for caAERS using
  caCORE SDK tools

3
Metadata Project
4
Metadata Project Areas Selected

• Selection criteria
• Classes that are in the top level models (BRIDG,
  HL7 RIM) as well as sub-models (caAERS)
• Classes that are stable
• Classes of interest to other sub-domain models

• Selections
• Organization
• Person
• Study Subject
• Product
• Location
• Drug
• Ingredient
• Observation
• Clinical Study
• Protocol
• Procedure
• Treatment

5
Metadata Project Work Approach Defined

• Create common representation taking into
  consideration
• Best features of both models (BRIDG caAERS)
• Other models and CDEs
• caBIG Standard data elements
• caCore SDK constraints
• Ease of understanding for both, models and
  generated data elements
• Adaptability for other sub-domain models

6
Metadata Project Simplified Representations

• Façade Design Pattern based generic approach
• The generic approach applied to Organization

7
Metadata Project Simplified Representation
(Generic Approach)
Reference Model
Implementation Model
Authentic Representation
Façade Representation
8
Metadata Project Simplified Generic Party Class
9
Metadata Project Simplified Generic Party Class
After
Before
10
Metadata Project Simplified Representation of
Organization
11
Common Data Elements for caAERS
12
CDE Generation using caCORE SDK Process
13
Lessons Learnt Modeling

• Avoid Complex Data Types
• Use the attributes of the simple data types
  instead
• Reduce use of extensibility feature of our
  classes
• Make attributes concrete instead
• Naming
• Post coordinated concepts as they exist in NCI
  Thesaurus
• Some suffixes for attribute names work better
  than others, like Type in contrast with Code

14
Lessons Learnt Tools

• Approximately 30 issues were pointed out to NCICB
  
• Bugs and Annoyances
• Ideas for Improvements

15
Lessons Learnt Tools - Bugs and Annoyances

• Examples
• Sometimes when the save command is given, the
  file on the disk is not updated. Workaround Use
  save as option and save a different copy of the
  file
• The original annotations for the UML entities
  (description, documentation, notes) get discarded
  in the annotated XMI file generated in the second
  semantic connector run.
• The semantic connector doesnt parse simple camel
  cased terms like administrationFrequency

16
Lessons Learnt Tools - Ideas for improvements

• Examples
• The concepts associated with an object class or
  property may be shown as leaves hanging off the
  branches from the navigation tree on the left.
  This can then be extended further by allowing
  dragging and dropping/copying of concepts from
  one entity to another
• One should be able to drag and drop concepts in
  the main panel to reorder them
• One should be able to create a local cache of
  frequently used concepts without having to
  connect to EVS every time a known concept is
  needed.