an easier way to prepare clinical trial data for reporting and analysis
TRANSCRIPT
Copyright BioPharm Systems, Inc. 2009. All rights reserved
An Easier Way to Prepare Clinical
Trial Data for Reporting and
Analysis
Mike Grossman
VP, Clinical Data Warehousing & Analytics
Thank you for joining. We will begin shortly.
Agenda
• Preparing Data for Analysis
• What is Study Data Mapper?
• Component Overview
• Other Key Features
• Questions and Answers
Agenda
• Preparing Data for Analysis
• What is Study Data Mapper?
• Component Overview
• Other Key Features
• Questions and Answers
Preparing Clinical Data for Analysis: Current State
• Companies collect clinical data primarily from EDC
systems that are supplemented with other sources, such
as central labs
• Many sponsors are increasing the use of CROs to conduct
trials
• To control costs and save on resources, companies are
preparing standard SDTM+ data models immediately after
the EDC extract process to be able to re-use analysis from
other studies
• The preparation of SDTM+ is still very labor intensive
Holistic Reference Clinical IT Reference Architecture
Outcomes
Common Data
Model
Project level
Conformed Data
Value Added
Study Data
Conformed Study
Data
Operational Trial
Metrics
Inbound
Data
Sources
Master Meta Data
AES & Complaints
Outcomes
External Study
Data
LIMS/PK
Central Labs
CDMS/ EDC
CTMS
Staging
Area
AES & Complaints
Source Specific
Outcomes Data
Shared Study and
Project Meta
Data
Study Specific
Data Staging
Trials
Management
Warehouse
Area
Specialized Data
Marts for
Scientific
Exploration and
Mining
Specialized Data
Marts for
Scientific
Exploration and
Mining
Specialized Data
Marts for
Scientific
Exploration and
Mining
Patient Sub
Setting and
Safety
Warehouse
Clinops Data
Marts
Meta Data Libraries, Version Control, Compliance Change Mgt
Ad-Hoc Query Dashboards Structured Reports Analytical Tools
Strategic
Analysis
Regulatory
Reporting
Data Mining
Clinical
Developmen
t Planning
Preparing SDTM+: Current State
• Preparing SDTM+ requires a mapping specification from
EDC and Labs, etc.
• Programming the transformation from raw data structures
to SDTM+ requires skilled programming resources
• Each study’s mapping specification requires a fresh start
with no clear strategy for tracking what can be re-used
from previous specification
• Subtle differences in source structures make it very difficult
to re-use standard code from one study to the next
Preparing SDTM+: A Better Way
• Preparing SDTM+ requires a mapping specification from
EDC and Labs, etc.
• SDTM+ standard is managed and tracked to evolve over
time
• The mapping specification is used to computer generate
the transformation code
• Re-use of mappings from one study to another is computer
assisted through metadata search
• The skill set for preparing mappings is changed from
programmer to data analyst
• BioPharm has built a software product called Study Data
Mapper as an example of implementing this new approach
Agenda
• Preparing Data for Analysis
• What is Study Data Mapper?
• Component Overview
• Other Key Features
• Questions and Answers
What is Study Data Mapper?
SDMapper is:
• A standalone software application for managing data
structures used for conducting clinical research
• An application for managing transformations between data
structures
• An application for intelligently re-using standards and
transformations
• An application for specifying standards and mappings
• Co-developed between ICON Clinical Research and
BioPharm Systems
What is Study Data Mapper?
SDMapper is:
• An integrated Excel spreadsheet template for specifying
and communicating metadata structures and
transformation maps
• An Oracle database for securely storing all the information
• A code generator for producing executable programs
based on the supplied metadata and mappings
• Includes features for version control, standards
management and validation management
• Designed specifically for clinical trials and other data
Study Data Mapper: Process Focused
1
Specify
metadata
and
mappings
2
Upload and
parse for
errors
3
Store under
version
control
4
Generate
executable
code and test
5
Recommend
re-use of
mappings
(TBD)
6
Download with
recommendations
7
Metadata Reports
Agenda
• Preparing Data for Analysis
• What is Study Data Mapper?
• Component Overview
• Other Key Features
• Questions and Answers
Mapping Project
• A mapping project is an object used to manage the
metadata for a set of tables and value lists. It may also
manage the relationship between that metadata and
transformations to other structures.
• Examples of a mapping project
– The list of tables and controlled terminology for the SDTM+ data
standard
– The definition of the source data structures from EDC and Labs and
their mapping to the CDISC SDTM
– The mapping specification for pooling tables form multiple studies
where the studies are similar but not the same.
• Key attributes of mapping projects
– Name, Version, Validation State, Standard or Non-standard
Mapping Project Example
Table Set
• An object that contains metadata for a group of uniquely
named tables
• Examples
– All the tables extracted from an EDC system for a study
– All the tables in the SDTM+ data standard
• There can be multiple table sets in a single mapping
project
• When mapping is being used, a table set may be identified
as a source or a target
– Typically, there will be one or more source table sets and a single
target
Table
• A table is a metadata description of a data table, similar to
a SAS dataset or a database table
• A table belongs to a Table Set
• Examples of a table
– DM, EX, AE from CDISC SDTM
– The data extract view structures from Oracle Clinical
– Tables from the pooled data model for a therapeutic area
• Selected metadata attributes of a table
– Name, Version, Validation Status
– Table Label, Table Order, Repeating, Source Location, Crfloc,
Crfnote, Odm Structure, Odm Repeating, Odm Is Reference Data,
Odm Purpose, Odm Class
Example of a Table Set and a Table
Variable
• A variable is a metadata description of a column of a table
• A variable usually belongs to a table but can belong to a
Table Set
• Examples of a variable
– USUBJID, SEVERITY
• Selected metadata attributes of a variable
– Name, Version, Validation Status
– Vpkey, Vorder, Vlabel, Vlabellong, Vtype, Vlength, Vprecision,
Vformat, Vformatflag, Vcase, Ismandatory, Istemporary,
Vimportance, Vderivetype, Vdatadomain, Sdtmflag, , Vheader,
Vcrfloc, Vcrfnote, Vrole, Vorigin, Vsuppqualflag
Value Lists
• A value list is a named ordered list of possible values
• A value list belongs to a table set. Typically a table set
includes all the value lists for the controlled terminology
associated that are used by the tables in the same table
set
• Examples of value lists
– CDISC code list for Severity
– Oracle Clinical DVGs
– Units for labs
• Value list attributes - Name, Description, Version, values
– Value list value attributes-Start value, End value, Code, Sort order
Example of a Value List
Map Sets, Table Maps, and Value List Maps
• A map set is a group pf table mappings that specify the
transformation rules from one or more table sets to a target
table set
– Examples
• Transform from an EDC and central lab structure to the sponsor data
standard
• Transform selected tables from multiple studies to a single set of
pooled data
• A map set belongs to a mapping project
• A table map maps from one or more source tables to a
target table
• A value list map maps values between two different value
lists
Table Map Use Case
MappingProject–SDTM
Metadata
MappingProject–STUDY123
MapSet–STUDY123toSDTM
TableMap(DM)
Sub-MapGroup
Sub-Map
TableMap(VS)
Sub-MapGroup(VS1)
Sub-Map(Common+HGT)
CommonMappings–SBP.DBP.HRT
TableMap(LB)
Sub-MapGroup
Sub-MapGroup
Sub-Map
Sub-Map(Common)
Sub-Map
Source Target
DM
VS
LB
Join
DEMO
VS001
IC
VS002
LAB
LABExt
Variable Mappings
• Given a single output variable, you can map a single
variable via an equals operator or any combination of
variables through a more complex operator
• Examples
– HUB.DM.USUBJID=EDC.DEM.PATIENT
– HUB.DM.USUBJID=CATX(‘-’,HUB.DEM.STUDY,HUB.DEM.PT)
• Any number and complexity of operators may be registered
based on your installation and needs
• Automatic data type conversions
• Automatic conversion of terminology if value list maps are
defined
Example Simple Variable Map
Example Value List Map
Example Common Mapping
Mapping Project After Upload to SDMapper
Code Generator
• Mappings defined can generate code in a few seconds
• Each table map is generated separately and called in the
order that takes care of dependencies
• Current version generates portable standalone SAS code
that has no dependencies on SDMapper or LSH.
– Ideal for internal use, as well as sharing with partners and
regulatory authorities.
• Future versions will generate SAS programs or PL/SSQL
programs directly to an LSH environment
Example Code Generation
Example Generated Code Snippit
Agenda
• Preparing Data for Analysis
• What is Study Data Mapper?
• Component Overview
• Other Key Features
• Questions and Answers
Other Key Features
• Recommendations engine: This is a future feature that will
take your existing source and target structure and look
through already existing mappings and recommend
starting with similar mappings if they already exist. This will
maximize re-use of existing mappings.
• Tagging: This is the ability to assign attributes to any
object. The current version lets you tag the sponsor and
the study for a mapping project. Future versions will allow
for any tags, such as compound, therapeutic area, concept
etc.
Agenda
• Preparing Data for Analysis
• What is Study Data Mapper?
• Component Overview
• Other Key Features
• Questions and Answers
Contact Information
If you have additional questions, please contact:
United States: +1 877 654 0033
United Kingdom: +44 (0) 1865 910200
Email Address: [email protected]
Website: www.biopharm.com