Standards Driven Innovation
Frederik Malfait
IMOS Consulting GmbH, Hoffmann-La Roche AG
PhUSE Annual Conference 2014
Managing Standards
2
Data Standards Value Proposition
• Standards are increasingly mandated – You have to do it anyway
– You can as well make the best of it
• Define once, use many times
• When submitting data and analysis results, standards can ensure – Compliance
– Consistency
– Correctness
• Have a repeatable process, do things faster
• Make data semantically meaningful – Precisely define models and content
– Store semantics with the data, so it can be remembered
– Supports understanding and transparency of the data
– Semantically align and integrate data from differen^t sources
• Standards based automation – Store data standards in a Metadata Registry
– Workflow automation
3
4
Pieces of the Puzzle
Standards Management
Standards Governance
Standards
MDR
Standards Automation
Content
Management
Information
Architecture
5
Models and Standards
M3: Meta-Meta-Model
M2: Meta-Model
M1: Model
M0: Model Runtime Data
CDISC
ISO 11179
RDF
6
ISO 11179 Standard for Metadata Registry (MDR)
7
RDF Meta-Model Schema
8
RDF CDISC Schema
9
Model Instantiation: SDTM
10
Example - SDTM
11
Example - SDTM
12
Example - SDTM
13
Example - SDTM
Roche Implementation
• Integrated data standards Roche and Genentech 2010
• MDR with operational metadata – Enterprise level application, fully validated system
– First production release 2011
• MDR components – Browser, web services, search, item level versioning
• MDR concepts – Prepared first half of 2013
– To be integrated in standards development over the next few weeks
• Model driven capabilities (2014-2015): use RDF to specify – Model driven read/write web servies + XSLT engine
– Model driven search
– Model driven validation rules
– Model driven UI
– Model driven security
• MDR used by other parties in Roche – Master data management
– Configuration of integrated document managment systems
14
PhUSE CSS Semantic Technology WG
• Existing CDISC Standards in RDF – CDASH
– SDTM
– SEND
– ADaM
– Controlled Terminology
• Protocol and Schedule of Activities in RDF
• Analysis Results Metadata in RDF
• Link to EHR – Co-lead from CDISC
• Regulatory Guidance in RDF – Co-lead from FDA
15
MDR Driven
Study Workflow Automation (SWA)
16
Clinical Data – Future State
17
External
Knowledge
Sources
Study
Design Protocol
Data
Analysis
Plan
Data
Collection
Design
Data
Collection
Data
Tabulation
Data
Analysis
Clinical Study
Report
Submission
Clinical
Development
Plan
Standards
& Metadata
Repository
Standards & Metadata Repository: Integrated Workflow
Building out the vision, step by step
18
Standards can help drive automation of key clinical processes…
Automate existing information flow
MDR
Experiments
EDC Build
SDTMv
Key Benefit
Information
Standards
Extension
Study
Workflow
Automation
Roadmap
SoA
Submission
Study
Design Protocol
Data
Collection
Design
Data
Collection
Data
Tabulation
Data
Analysis
Clinical Study
Report
Standards
& Metadata
Repository
Data
Analysis
Plan
Clinical
Development
Plan
SWA Experiments Status
19
Experiment Description Status
Exp. 1 Create a schema for a machine-readable SOA. Complete.
Exp. 2 Upload a SOA into the GDSR. Complete. Includes representation
of SOA in Word and HTML.
Experiment Set A: Creating a machine-readable SOA
Exp. 3 Use the GDSR to generate an operational annotated eCRF
of the Global Data Standards.
Complete.
Exp. 4 Use the GDSR to generate submission-ready annotated
eCRFs of the Global Data Standards.
Complete.
Exp. 5 Use the GDSR to generate ALS templates including edit
checks of the Global Data Collection Standards.
Complete (excluding edit checks).
Exp. 6 Develop a tool that will produce a mapping specification
and corresponding programming environment-independent
SAS program that transforms collected data to SDTMv.
Ongoing.
- Code list alignment opportunity
identified.
- Reuse of data element mappings
across forms
- Potential to generate SDTMv
annotations from mappings.
- Platform-independent
Experiment Set B: Leveraging Global Data Standards
SWA Experiments Status
20
Experiment Description Status
Exp. 7 Design an interface that will enable the
definition eCRFs from Global Data Standards
or definition of new eCRFs.
First SME workshop completed.
Process analysis ongoing.
Exp. 8 Design an interface that will enable the
definition a non-CRF File Format
Specification.
First SME workshop completed.
Process analysis ongoing.
Exp. 9 Design an interface that will enable the
definition of a Visit Form Matrix.
First SME workshop completed.
Process analysis ongoing.
Exp. 10 Use the GDSR to generate a study-specific
ALS template from study-level metadata.
First SME workshop completed.
Process analysis ongoing.
Exp. 11 Use the GDSR to generate a non-CRF File
Format Specification from study-level
metadata.
First SME workshop completed.
Process analysis ongoing.
Experiment Set C: Expedite the Study Build
Schedule of Activities - SDM XML
21
Schedule of Activities
22 Schedule of Activities
Web
Service
Operational CRF and Submission CRF
23
Data
Collection
Web
Service
Rave Architect Loader Spreadsheet
24
Data
Collection
Web
Service
Edit Checks
25
Data
Collection
SDTM Transformations
26
Data
Collection
Web
Service
Data
Tabulation
Source
Data Target Data
Mappings
SDTM Mappings
SDTM Transformations
• Reuse of existing data standards – Rave source data elements
– SDTM target data elements
• Generate easy to understand SAS code
• Maximize platform independent code
• Support multiple target platforms
• Possible extensions – Generate mapping specifications
– Generate conformance checks
27
Schedule of Activities
28
Arms and Epochs
29
Planned Activites
30
Time Points and Timing
31
SoA Instance
32
SoA Instance
33
Findings
• Full protocol too large to attack at once
• Schedule of Activities key for downstream automation
• Move data collection standards from forms to modules, research concepts, and
data elements
• Hard, but necessary for first release – Handle protocol amendments
– Integrate with data standard request process for new items
– Enterprise integration (App, MDR, EDC, data platform, CTMS etc,)
• Business value – Less on the protocol side
– Massive on automated EDC build
– Massive on automated SDTM transformations
• Later stages – Link SoA to cost information
– Integrate endpoints and data analysis standards
– Component based authoring
34
Protocol and SoA Concepts
• TransCelerate Common Protocol Template – Human readable protocol
– Endpoints
– Technology sub-team
• CDISC Protocol Concept List – Excel list, need to accellerate progress
– Started with concept comparisons in different settings
• PhUSE – Prepared first half of 2013
– Integrated in standards development, second half of 2014
• All these activities complement each other
• Concept Management is a key component to be successfull
• End to end standards – Create CDISC standards for TA endpoints and activities
– Link activities to data collection collection standards and metadata
– Link endpoints to data analysis standards and metadata
35
36
Questions ?