HealthInfoCDA: Case Composition Using Electronic Health Record Data Sources

1
HealthInfoCDA Case Composition Using Electronic
Health Record Data Sources

• Grace I. Paterson, Syed Sibte Raza Abidi
• and Steven D. Soroka

2
Guiding Principle Leverage standards associated
with Health Level 7 (HL7) Clinical Document
Architecture (CDA) to develop an infostructure
for clinical care and education
3
Goal

• Capture clinical experience in a case-based
  reasoning (CBR) system from
• Input 1 electronic discharge summaries
• Input 2 longitudinal electronic health records
• Input 3 guidelines and online handbooks
• Evaluate the HL7 Clinical Document Architecture
  (CDA) model as a boundary object that
• represents clinical activity
• supports information flow
• bridges practice and theory through a CBR system

4
Creation of a CDA Document

• Separate header and body data in source
• Identify sections
• Identify clinical statements in each section
• Identify entries (ActEncounterObservation
  ProcedureSubstanceAdministration...) in each
  clinical statement
• Identify codes or code-sets for each entry

5
Electronic Discharge Summary

• Objective produce discharge summaries that are
  more complete and contain more of the essential
  data elements
• Method create CDA document--Chronic Kidney
  Disease Discharge Summaryusing a Template
• Subjects Nephrologists, Residents, Nurse
  Practitioners (N5)
• Results all beta-testers found the amount of
  content was appropriate, but requested
  improvements in layout and interface design.

6
Lessons Learned

• Variation in what Communities of Practice choose
  to observe
• Specialists are most parsimonious
• Record-keeping skills are not explicitly taught
• Template layout should accommodate different
  charting styles
• Discharge Diagnosis is common to all
• Health Status is inconsistent because it is
  subjective
• Concept Coding requires attention to semantic
  classes

7
Longitudinal Health Care Record

• Objective Identify and encode clinical
  statements from documents in patient chart
• Method
• identify concepts in text using natural language
  processing tools and encode to SNOMED UMLS
• transform paper chart to CDA documents
• use HL7 Clinical Statement Pattern in Sept 2005
  HL7 Ballot Package for document construction
• Result structures that retain care context and
  support case composition

8
Guidelines and Online Handbooks

• Objective Aggregate concepts into knowledge
  elements for case-based teaching
• Method
• identify concepts in text using natural language
  processing tools and encode to SNOMED UMLS
• produce knowledge elements for concepts
• aggregate knowledge elements using clinical
  action reference model
• Results A method for achieving semantic
  interoperability between Observations and Actions
  in CDA and Policies and Decision knowledge
  resource

9
Observations
Communities of Practice
Policies
SNOMED
HL7 CDA
Collaboration
Communication
Structure of a Document
Meaning of a Concept
Knowledge Transfer
Actions
Decisions
10
Case-Based Reasoning

• Structural mapping between CDA-based electronic
  health record and Case
• Problem/Situation Description history,
  observations
• Solution clinician inferences in assessment
• Outcome resulting state expressed as followup
  instructions to next clinician
• SNOMED as switching language between concepts in
  case base and knowledge sources
• Produce learning objects for chronic disease
  management activitiescondition tracking,
  condition assessment, care planning

11
Learning Object Example