The caBIG/HL7/CDISC Structured Protocol Representation

1
The caBIG/HL7/CDISC Structured Protocol
Representation
John Speakman
Douglas B. Fridsma, MD, PhD Center for Biomedical
Informatics Center for Pathology and Oncology
Informatics University of Pittsburgh
2
Agenda for today

• A short history of the model and what it does
• (aka caSPR, aka Problem Space Model, aka DAM, aka
  DSAM)
• CaPRI and the Development Framework
• How this workspace could/should get involved

3
A Team Effort

• Core modeling team
• Christo Andonyadis (NCICB)
• Greg Anglin (Eli Lilly)
• Lisa Chatterjee (Digital Infuzion / NIAID)
• Julie Evans (CDISC)
• Doug Fridsma (UPCI)
• Smita Hastak (ScenPro / NCICB)
• Ray Heimbuch (Novartis)
• Charlie Mead (Oracle / HL7)
• Joyce Niland (City of Hope)
• John Speakman (Memorial Sloan-Kettering)
• Cara Willoughby (Eli Lilly)
• Diane Wold (Glaxo SmithKline)

Leadership Sue Dubman (NCI), Becky Kush (CDISC),
Charlie Mead (HL7 RCRIM / Oracle), Linda Quade
(HL7 RCRIM / Eli Lilly), Barbara Tardiff (HL7
RCRIM / Merck), Randy Levin (HL7 RCRIM / FDA)
4
Clinical Data Interchange Consortium
(www.cdisc.org)

• Protocols
• SEND

Operational Data Interchange Archive ODM, LAB
Submission Data Interchange Archive SDM, ADaM

• Data Sources
• Site CRFs
• Laboratories
• Contract
• Research
• Organizations
• Development
• Partners

• Operational
• Database
• Study Data
• Audit Trail
• Metadata

• Submission
• Data
• CRT/Domain Datasets
• Analysis Datasets
• Metadata

One harmonized CDISC standard/model
ODM Operational Data Model SDM Submission
Data Model LAB Laboratory Data Model ADaM
Analysis Data Models
5
Operational Data Modeling
Submission Data Modeling
Regulatory Reviewers
ODM values can be expressed via SDS metadata
Data Sources

• ODM
• Operational Data Model
• XML format for clinical study
• data and metadata
• Supports regulatory-compliant
• e-Archive format

• SDS
• Submission Data Standards
• Describes content, structure and attributes of
• data for regulatory submissions
• Defines metadata content for standard domains
• Facilitates regulatory reviews

define.xml

• LAB
• Laboratory
• Data Model
• Defines standard content for
• transferring clinical laboratory
• data from labs to sponsors
• ASCII, SAS, XML, and
• HL7 implementations available

LAB data can be mapped into ODM XML
Applies submission model for analysis datasets

• ADaM
• Analysis
• Dataset Models
• Provides models for commonly used statistical
  analyses.

ODM can be used to represent and archive
submission data
6
Health Level Seven (www.hl7.org)

• The worlds leading standard for the electronic
  interchange of healthcare information
• gt20 Global affiliates
• 16 years of operation
• ANSI-accredited SDO, acknowledged by the DHHS as
  the standard for healthcare information exchange
• RIM (Reference Information Model)
• RCRIM (Regulated Clinical Research Information
  Management) Technical Committee

7
(No Transcript)
8
CDISC and HL7 Relationship

• 2001-2002
• Early CDISC attempts to map CDISC ODM
  (Operational Data Model) to the HL7 RIM failed
• 2003
• CDISC-HL7 joint study group to assess how CDISC
  standards might become RIM compliant and be
  grafted onto HL7 RIM
• Plan developed for creating domain-space
  business model for clinical research
• 2004 Modeling starts

9
Domain Space Analysis Model Timeline
2001-2002
Jan 2004
Mar-Aug 2004
Aug-Oct 2004
Dec 2004
2003
Nov 2004
...
Dr. Meads Report to CDISC Board of Directors
Initial DSAM Development, led by Dr.Mead

Feedback Review of DSAM with Focus Groups
Begin Mapping CDISC ODM to HL7 V3 RIM
Initial Modeling of Protocol Elements in DSAM
Early attempts to map ODM to HL7 RIM

• Involvement from
• CDISC
• NCI
• RCRIM inv.

• Reps from
• NIH/NCI
• FDA
• EHR/eSource
• TCC
• HL7 RCRIM
• CDISC teams
• CDISC IAB

• Reps from
• CDISC Board
• CDISC Teams
• HL7 RCRIM
• TCC
• NCI

• Reps from
• CDISC
• HL7

• Reps from
• CDISC
• NCI/caBIG
• HL7 RCRIM

HL7 V3 RIM Implementation of CDISC LAB Model -gt
ANSI Accreditation Assistance for ODM sought
from L.Bain, C.Mead
SDTM Announced by HHS/FDA eSubmission made w/
define.xml
ODM Production V1.2 (XML Schema) - Industry
Adoption and Tool Development ?
10
Interchange Standards Long-term Desired
Outcomes

• A holistic approach to standards, facilitating
  data interchange from sites through regulatory
  submission, utilizing XML
• Standards for data acquisition supporting the
  population of a cross-trial warehouse within FDA
• HL7-CDISC models harmonized to yield value for
  both clinical research and healthcare sharing
  of information between EMR and clinical trials
• Global adoption of CDISC data standards

CDISC Meeting with FDA Commissioner, April 2003
11
CDISC Plans

• Today, CDISC is working to ensure
  interoperability among its own standards.
• CDISC has committed to using the HL7 Development
  Framework (HDF) as a process through which the
  CDISC standards are to be harmonized with the HL7
  RIM in the long term.
• The ultimate goal is one overarching model to
  support interoperability between healthcare and
  clinical research systems.

12
The World of Clinical Trial Standards
Source Becky Kush, mod.Cara Willoughby
TC RCRIM
Protocol Std
Dictionary, Codelist
Organization
Document Standard, or Architecture
Standard
Model
13
Vision of the SPR SIG (July 2004)

• To
• Define a computable protocol representation that
  supports the entire life-cycle of a clinical
  trial protocol
• Develop tools that use this protocol
  representation.
• The protocol representation will serve as a
  foundation for caBIG modules that support all
  phases of the clinical trials life cycle
  (including protocol authoring)

14
Todays Clinical Trial Protocol
One full-text source presented several ways
Source Greg Anglin, Eli Lilly
15
Tomorrows Clinical Trial Protocol
A data layer in the full-text source
Data layer
16
Interchange vs. Interoperability

• interoperability ability of two or more
  systems or components toexchange information and
  to predictably use the information that has been
  exchanged. Source IEEE Standard Computer
  Dictionary A Compilation of IEEE Standard
  Computer Glossaries, IEEE, 1990

Syntax ? Structure Semantics ? Meaning
Source Charlie Mead, HL7
17
The Semiotic TriangleHow Humans Communicate
Source Charlie Mead, HL7
18
Protocol and the Semiotic Triangle
Symbol Protocol
19
Some HL7 Terms

• Mood
• Lifecycle of a concept through a business process
• State
• Lifecycle of an instance

20

Study
Concept
Propose
Plan
Execute
Report
Business Process
Protocol
Document
Instance
21
ProtocolA document that contains study
information in the planning mood.
Intent
Study
Report
Document
Plan
Execute
MOODLifecycle of a research concept through a
business process
STATE Lifecycle of an instance
22
Basis of the Model

• CDISC Protocol Elements Spreadsheet
• Contains definitions of 200 concepts deemed
  important to clinical research
• But no structure (i.e., no representation of
  their relationships)
• Job 1 build a model that includes all the
  concepts in the speadsheet

23
The iterative process of creating the SPR Stage I
HL7v3 RIM
Domain elements (drawn from CDISC PR elements)
at this stage there are granular ideas that we
know we need to represent, but relationships not
clear
Business Inputs
24
CDISC protocol elements
25
The iterative process of creating the SPR Stage
II
HL7v3 RIM
Strawman model of structure of domain elements
at this stage we describe the relationships
between domain elements, without regard for the
RIM
Business Inputs
26
(No Transcript)
27
(No Transcript)
28
The iterative process of creating the SPR Stage
III now
HL7v3 RIM
reductionist model of structure of domain
elements
at this stage we analyze the model to boil it
down to its primitive data structure. It gets
close to RIM, but very abstract loses some
meaning to domain experts
Business Inputs
29
The Model Stage II-III
30
The iterative process of creating the SPRStage
Two ? Stage Three ? Stage Four
31
The iterative process of creating the SPR Stage
IV
RIM
integrated model(s) of structure of domain
elements
Explode back to domain concepts, through
instance diagrams, storyboards,vocabularies, and
other tools check that the initial elements
still hold in the model
Business Inputs
32
ECOG 2190 Schema
33
ECOG 2190 Protocol Activity
34
caSPR forms the foundation for interoperable
module development
35
Proposed caPRI System Description
36
Portions currently implemented
EAV database
37
(No Transcript)
38
(No Transcript)
39
The Importance of Process

• We need processes to develop interoperable
  applications
• Can use the HL7 Development framework
• We need processes to develop interchange
  standards
• Again can use the HL7 Development Framework
• We need processes to maintain the Model
• Collaboration
• Versioning control
• Model harmonization

40
A example development processcaPRI
Application Development
caSPR module analysis model
caSPR CDE
caPRI input module (module specification)
caPRI CT management framework
Functional Requirements
caDSR candidate CDEs
Semantic connector UML loader
caSPR CDISC DSAM HL7 DAM
RCRIM DMIM
RMIM (interchange requirements)
HL7 XSD (Interchange specification)
HL7 compliant framework
Interchange Specification
Increasing specificity and computability
41
The people processes Desiderata for a
collaboration web-board-type thing

• Open access, open communication, discussion
  forums
• Mechanisms for proposing changes
• Mechanisms for harmonization
• Levels of access (readers, modelers, stewards)
• Tightly controlled versioning with tracking
• Integration of model packages, application code,
  documentation, and user groups

42
What have we accomplished?

• caSPR
• Established excellent collaboration with CDISC,
  HL7 and other caBIG modelers
• Constructed the initial pieces of a comprehensive
  model
• still much to do
• Established this model as the HL7 Domain analysis
  model
• Have established Enterprise Architect as the tool
  of choice, and moved the model from Rational Rose
  XDE into EA

43
What have we accomplished?

• caPRI
• Demonstrated a caBIG compliant application that
  is modular, extensible, flexible, and driven from
  the SPR model
• Set the stage for collaborative development of
  new modules and features
• Process
• Proposed a model development, collaboration, and
  version control process

44
The caFuture
45
What we need

• Ask not what the structured protocol
  representation can do for you,
• Ask what you can do for the structured protocol
  representation

46
Current caBIG/HL7/CDISC SPR needs

• Developing Instance diagrams, examples, and
  testing the model using the ECOG 2190 protocol as
  our initial test
• The model is static heavy
• Develop the behavioral portions of the model
• The people processes around clinical trials
• Calendars
• Activities
• Behaviors
• Study execution

47
Next steps in the SPR project our tasks

• caSPR
• Dissemination and presentationroad shows
• caBIG April 10th, HL7 May 2nd, CDISC middle/end
  of May
• Development and implementation of model
  versioning control and collaboration processes
• People and technical processes to be finalized
• Incorporation of informal comments and correction
  of known model problems
• Next core modeling session next week!

48
Next steps in the SPR project

• Model development needs
• Review class and attribute definitions
• Assistance with instance diagrams and examples
• Requirements gathering for new functionality and
  subdomains
• Application development needs
• Adopters of caPRI testing and improvement
• Developers of new modules

49
Next steps in the SPR project

• Integration within CTMS and across caBIG
• SIGs should be the leads in the requirements and
  testing of the parts of the model that they have
  expertise in
• Other workspaces activities will touch on the SPR
• Tissue banking and pathology tools
• i.e., us

50
TBPT involvement

• Include a stub to the caTISSUE OM?
• How big is the stub?
• Or just include the whole caTISSUE OM within it?
• When is a tissue protocol something we care
  about?
• When is it just another protocol like the IRB
  protocols we know and love?
• What does it depend on?
• Consent type?
• IRB involvement?
• Assignment of an IRB number?
• Annotation of specimens?
• shrug