presentation

Download Report

Transcript presentation

An Open Source Reporting System
for Life Sciences
Magnus Mengelbier
Director
PhUSE 2010
1
Topics







PhUSE 2010
Introduction
Traditional process
Library
Analysis
Reporting
Connections
Summary
2
Tradition
Listings
Source
Derived
Tables
Graphs
Source
Listings
SDTMs
ADaMs
Tables
Graphs
Define
PhUSE 2010
3
Templates and Specifications
Specifications
SAP
Shells
Shells
Data
Sets
Data
Sets
Programs
Templates
Data Sets
Tables
Listings
Define
SAP
PhUSE 2010
4
Templates et al
PhUSE 2010
5
Library
Specifications
Shell
TLGs
Data
Sets
Data
Sets
Programs
Library
Data Sets
Tables
Listings
Define
SAP
PhUSE 2010
6
Organizing Content





Directory Trees
No restriction on the number of
levels
No restriction on naming
conventions
Contexts
Aware of parent directory
 Contexts have to be consistent within a directory tree
 Directory tree can be your departmental drive or network share
PhUSE 2010
7
Default Defintions

Data Sets
 Tables
 Listings
 Graphs
PhUSE 2010
8
Definitions and Inheritance


Types of definition controlled for each level
Definitions can be inherited from parent to child levels
Report
x
Study
x
x
x
x
x
x
Indication
Project
Global
PhUSE 2010
x
x
STUDYID
AGE
x
FEDFAST
Data Set
AE
Data Set
XY
9
Specifications
Specification
Library


PhUSE 2010
A Specification is just a set of links to content in the Library
All Definitions have to be in the Library
10
A Comparison of Defintions
PhUSE 2010
11
Repository
Data Sets
Tables
Listings
Graphs
Simple Meta-Data Engine


PhUSE 2010
Model – View – Controller (MVC) design pattern
 Model – Meta-data
 View – Graphical and programming interface
 Controller – ”Administrator” of the meta-data
Very primitive and flexible meta-data engine
12
Repository - Definitions

Defintions
 Type, e.g Data Set, Table, etc.
 Identified by Type and Name

References and associations to other meta-data items
 Look-up
 Ordered list

Versions
 Increment
 Construct for a unique reference to variations of the same
item
 More than one version is selectable
 Future plans include major/minor versions
PhUSE 2010
13
Repository – Data Set
Attribute
Value
Name
CLASS
Label
Class
Sort order
Key sequence
PhUSE 2010
Sequence
Reference
1
NAME
2
SEX
3
AGE
4
HEIGH
5
WEIGHT
Collections
14
Repository – Table (1)
PhUSE 2010
15
Repository – Table (2)
Attribute
Value
Name
AGE
Label
Age (years)
Link
DM.AGE
PhUSE 2010
Attribute
Value
Attribute
Value
Name
N
Name
MEAN_SD
Label
N
Label
Mean (SD)
Format
[2]
Template
MEAN (SD)
Decimals
[0]
16
Repository – Table (3)




N
MEAN_SD
MEDIAN
MIN-MAX








PhUSE 2010
N
MEAN_SD
MEDIAN
MIN-MAX
MEAN
SD
MIN
MAX
17
Analytics
 Meta-data driven
 Standard statistics
 Statistics “meta-data” interface
 Requires an analytics engine (SAS/R/S-Plus)
 Code library
 Code generators
PhUSE 2010
18
Analytics
Library
Specifications
Code Generator
Code
TLGs
Data
Sets
Data Sets
Data Sets
Tables
Listings
Define
Non-standard
Programs
SAP
PhUSE 2010

N

MEAN_SD

MEAN

MEDIAN

SD

MIN-MAX

MIN

MAX
19
Reporting By Elements
Reporting
Analytics
Library

PhUSE 2010
Data
 Input Data Set
 Select Parameter(s)
 Subset
 Summaries & Statistics
 Compute
 Append
 Format
 Render Table
 Report structure
 Format
 Sort & Order
 Render
20
Reporting by Elements
Analysis Data Sets
ADAE
ADLB
ADVS
…
STAE
STLB
STVS
…
Statistics Data Sets
STSUBSET
Outputs
PhUSE 2010
21
Analytics Repository
Code Generator
Code Library
Non-standard
Programs
PK Analytics
Analytics Repository
PhUSE 2010
22
Reporting Results
Code Generator
Code Library
Non-standard
Programs
PK Analytics
Analytics Repository
Tables
PhUSE 2010
Listings
Graphs
Portals
Submission files
23
Other Outputs
Data Sets
Tables
Listings
Graphs
Simple Meta-Data Engine



PhUSE 2010
Model – View – Controller (MVC) design pattern
 Model – Meta-data
 View – Graphical and programming interface
 Controller – ”Administrator” of the meta-data
Very primitive and flexible meta-data engine
Adding an output starts with a Model, View and Controller(s)
24
Tracking and Status
Library
Code
TLGs
Data
Sets
Specifications
Code Generator
Data Sets
Non-standard
Programs
Data Sets
Tables
Listings
Define
Common Work Flow
 Track status
 Track and enforce dependencies
PhUSE 2010
25
Organization Revisited





Directory Trees
No restriction on the number of
levels
No restriction on naming
conventions
Contexts
Aware of parent directory
 Contexts have to be consistent within a directory tree
 Directory tree can be your departmental drive or network share
PhUSE 2010
26
Connections
Library
Specifications
Code Generator
Code
TLGs
Data
Sets
Data Sets
Non-standard
Programs
Data Sets
Tables
Listings
Define
Common Work Flow
Network
Subversion
SAS Server
SDD *
R
* SAS Drug Development
PhUSE 2010
27
Web Services
Library
Specifications
Code Generator
Code
TLGs
Data
Sets
Data Sets
Non-standard
Programs
Data Sets
Tables
Listings
Define
Common Work Flow
PC SAS
S-Plus / R
SAS Server
SDD *
Anything really
* SAS Drug Development
PhUSE 2010
28
Conclusion
Data Sets
Tables
Listings
Graphs
Simple Meta-Data Engine



PhUSE 2010
Model – View – Controller (MVC) design pattern
 Model – Meta-data
 View – Graphical and programming interface
 Controller – ”Administrator” of the meta-data
Very primitive and flexible meta-data engine
Adding an output starts with a Model, View and Controller(s)
29
Conclusion
Library
Specifications
Code Generator
Code
TLGs
Data
Sets
Data Sets
Non-standard
Programs
Data Sets
Tables
Listings
Define
Common Work Flow
Network
Subversion
SAS
SDD *
S-Plus / R
* SAS Drug Development
PhUSE 2010
30
Questions & Comments







PhUSE 2010
Introduction
Traditional process
Library
Analysis
Reporting
Connections
Summary
Magnus Mengelbier
Director
Limelogic Ltd
[email protected]
+44 208 144 5701
31
THE
PhUSE 2010
END
32