The presentation discusses how cognitive sciences and next generation clinical data management can transform clinical trials. It notes that currently, 72% of studies are one month behind schedule, 70% experience patient enrollment delays, and 20% do not recruit any subjects. It advocates centralizing and contextualizing data in a clinical data lake to enable evidence generation and reduce time and costs. The presentation outlines Saama Technologies' clinical data-as-a-service solution which uses metadata-driven transformation, analytics applications, and data pipelines to generate insights from varied data sources in real time. It argues that disruptive thinking is now required to achieve clean, longitudinal data and operational efficiencies through cognitive systems and a patient-centric, "Silicon Valley" mindset
1. Delivering Analytics Advantage
to the Global 2000
Next Gen Clinical Data Sciences
The time for Transformation is NOW…
Karim Damji
SVP, Products and Marketing
April 24, 2017
2. 222
AGENDA
Analytics Advantage
The Fourth Industrial Revolution
Cognitive Sciences in Medicine
State of Clinical Trials Today
Centralize & Contextualize Data
Next Gen Clinical Data Mgmt
Putting it Together
Speed & Specificity
01
02
03
04
07
05
06
10. 1010
Current state of Clinical Trials
Period: 2012-2016
Top-10 Contributors – All
1. Insufficient/low/poor enrollment
2. Lack of efficacy
3. Change in business focus (Sponsor)
4. Serious Adverse Events (SAEs)/death/safety
concerns
5. Investigational New Drug
withdrawn/discontinued
6. Protocol deviation
7. Terminated by regulatory agency
8. Funding support withdrawn
9. Administrative/ Technical reason/ Business
decision
10. Unavailable study personnel
Laggards
(20%)
Perfect
(10%)
Costly
(40%)
Over
Time
Help Me
(20%)
Abandoned
(10%)
On
Time
On
Budget
Over
Budget
11. 111111
72%
studies are
1 mo. behind
SCHEDULE
70%
of studies
experience
patient
enrollment delays
20%
does not recruit a
single subject
30%
of time spent on
patient
recruitment
14%
Trials are
abandoned
12. 121212
Discovery Pre-Clinical Phase 1 Phase 2 Phase 3 FDA Post Market
• Siloed Data
• Not Standard & Bad Quality
• Timeliness
• Scalability & Cost
But, there is a Data Problem
• Study Design & simulation
• Manage Multiple CRO’s
• Ongoing Risk – Enrollment, Study completion
• Operational Issues – Performance {Site, Study, Trial}
Trial Management has to stay on top
• Completed On Time
• Within Budget
• Detect failure early
Trials need to
Need for an Integrated
Data SolutionClinical Trials
• Effectiveness and Usage in Real World
• Comparative evidence against competitors
• Safety concerns & Adverse events
• Clinical & Economic value
• Any unmet needs ?
• Drug repositioning • Study design & simulation
• Clinical development feasibility
• Variety of Data (Avg 40-50 sources)
• Geography adds to complexity (EU/APAC different than US)
• Siloed & bad quality data
• Extremely expensive & time consuming considering data volumes (Weeks
and M$’s)
Need for an Integrated
Data Solution
13. 1313
Let’s just take 3 contributing factors…
Patient Matching
Site Selection
Comparative Studies
14. 1414
Patient Recruitment - Data Enrichment & Prediction
Target
Cohort
Historical Data
Social Media
Patient Clusters,
Competitive Studies
Trial safety data,
Seasonality
Age, Race, Economic data
High
Probable
Matches
Stack Rank with
related prior
Studies
Low
Probable
Matches
‘Predict’ the likelihood of the required recruitment
numbers and the most optimal match to make the
study a success in time and budget.
Model
For every
instance:
y =
β0+x1*β1+x2*β2
… x10*β10
15. 1515
Modeling Site & PIs
Site & PI
Probability
Model
Relevant Experience
Past success/failure
rates
Competitive Studies
Target Study Parameters
Scores
Model
For every
instance:
y =
β0+x1*β1+x2*β2
… x10*β10
16. 1616
• 2 Drugs
(1 off patent, 1 on market)
• Translational study
• Need differentiation
• Study characterize & pool
(~4000 subjects)
• Statistical Analyses
(Uni/Multivariate)
• Iterative functional review of
predicted probabilities
• Proven differentiation for 2
indications
• Subject characterization
• Repeatable data management
Use Case: Drug Differentiation
17. 171717
Clinical Data management
is ripe for disruption
60%Of trial activities
are data management
related
20%of trial $ for
clinical data sciences
$127B
2016 R&D
Spend
Total Market Spend
$148B
2020 R&D
Spend5-6%
18. 181818
AGENDA
Analytics Advantage
The Fourth Industrial Revolution
Cognitive Sciences in Medicine
State of Clinical Trials Today
Centralize & Contextualize Data
Next Gen Clinical Data Mgmt
Putting it Together
Speed & Specificity
01
02
03
04
07
05
06
19. 191919
• Proactive
intervention
• Strategic vendor
collaboration
• Financial
forecasting
Operation
• Track progress to endpoints
• Go/no go on next study
• In-sillico meta-analysis
Innovation
• Ingestion
• Quality check
• Standards
• Transform
• KPI’s
• Biostatistics
• Validation
Automation
Objectives & Targeted Outcomes
Data Mgmt. &
Quality
Centralized
Clinical
Data Lake
Evidence
Generation
Time to Insight
& Cost
20. 202020
Genomics
Internal, M&A, External,
Syndicated
Wearable Devices
High
Variety, Volume, &
Velocity Data
Analysis
Organization
Ingestion
Automated
Data Wrangling & Deep Data Science
Business Aware
Data Analytics Services Oriented
Architecture
Harmonization
Integration
Analysis
Organized Storage
Provisioning
Aggregation
Modern Technologies
Configurable
Analytic
Applications
Business
Outcomes
Approach
21. 212121
Ingest & Rules
Apply
Standardize
(CODM)
Apply Metrics
Analytic Ready
Data
Clinical Operations
KPI/Metrics Pipeline
Ingest & Rules
Apply
Standardize
SDTM
Convert to
ADaM
Analytic Ready
Data
Clinical Sciences
Subject Analytics Pipeline
CTMS
Project Manage
CRO
EDC
Labs
Biomarker
Clinical
Development
Analytics
Centralize (and contextualize) data in flight
24. 242424
Clinical Data-as-a-Service (CDaaS)
Enabled Analytics
Patient & Studies
Analytics
Clinical Study Data
Mart
Clinical Outcomes
Analytics
Drug Safety & Analytics
Safety Outcome &
Reporting Analytics
Trial Management
Analytics
Real World Signal
Detection Analytics
Activity Enablement
CDISC, MCC, Transcelerate Standards
Big DataRelational Data
Varied Structure Data
CDaaS
Advanced Analytical Tools
Meta Data Repository
Clinical Data
Safety Data
Varied Sources
Syndicated & Large Data
Data Sources
Electronic Data Capture
Clinical Trials Management
System
Safety Data Warehouse
Global Safety Data Warehouse
ARGUS
Clinical Study Reports
Disparate Business Unit Reports
External analyses
Non-Clinical, Pre-Clinical Data &
Reports
Real World Claims Data
Internal Genomics Data
Public Data (Kegg, NCBI,CHEMBL,etc.)
Trials Trove, CT.gov
25. 2525
On Going
Study
Workflow
• Mappings
• Rules
MDTS
1
MDTS
2
MDTS
3
Pattern
Match w/
Repo
Parameter
Generate
MSM ODM
Workflow
Generate
Source + ODM
Dependent
Metadata Driven Transformation Service
Rules – Based Audit Service
Data
Access
API
By By By
S
t
u
d
y
P
r
o
g
r
a
m
B
i
z
U
n
i
t
Vendor Cloud
Internal
• EDC
• Labs
• ePro
• IvRS
• CTMS
• Sample
Mgmt
• STDM
Patient Level
Clinical
Patient Level
Reported
eCRF +
Protocol
Legacy
Extracts
Master
Study
Metadata
Therapeutic Area
Phase
Drug Type
Version
Protocol
Study Event
Form
Item Group
Item
ODM Study
Metadata
Dictionaries
CRF
Protocol
Repo
CMT
High
Level
Rules
• DQ
• Validate
• Compliance
Metadata
Repository
Administrative
Technical
Descriptive
ODM
XML
Parse
Non-ODM
Recognize
Parse
Source > ODM
Mapping
Store
Clinical
Mappings
Data Entity
Look ups
Operations
Mappings
Data Entity
Look ups
CMT
Updates
Curation
Interface
1. CDR
2. PDR
1. CDR
2. PDR
1 2
SDTM Version
Resolution
Unified STDM
Version
Metadata Driven
Transformation
26. 2626
On Going
Study
Workflow
• Mappings
• Rules
MDTS
1
MDTS
2
MDTS
3
Pattern
Match w/
Repo
Parameter
Generate
MSM ODM
Workflow
Generate
Source + ODM
Dependent
Metadata Driven Transformation Service
Rules – Based Audit Service
Data
Access
API
By By By
S
t
u
d
y
P
r
o
g
r
a
m
B
i
z
U
n
i
t
Vendor Cloud
Internal
• EDC
• Labs
• ePro
• IvRS
• CTMS
• Sample
Mgmt
• STDM
Patient Level
Clinical
Patient Level
Reported
eCRF +
Protocol
Legacy
Extracts
Master
Study
Metadata
Therapeutic Area
Phase
Drug Type
Version
Protocol
Study Event
Form
Item Group
Item
ODM Study
Metadata
Dictionaries
CRF
Protocol
Repo
CMT
High
Level
Rules
• DQ
• Validate
• Compliance
Metadata
Repository
Administrative
Technical
Descriptive
ODM
XML
Parse
Non-ODM
Recognize
Parse
Source > ODM
Mapping
Store
Clinical
Mappings
Data Entity
Look ups
Operations
Mappings
Data Entity
Look ups
CMT
Updates
Curation
Interface
1. CDR
2. PDR
1. CDR
2. PDR
1 2
SDTM Version
Resolution
Unified STDM
Version
Metadata Driven
Transformation
Parse
Recognize
Meta Data Repository
Meta Data
Transformation Service
27. 2727
Unstruc
tured
Data
Unstructured Data
Structured Data
Syndicated Data
Transactional
System
Social Media Data
Streaming Data
API
Source
Data
Unstructured Data
Operating Systems
Windows Linux
On - Premise
Cloud
AWS MS AZURE Google Cloud
Hadoop Distributions
Amazon EMR
Unstructured
Data
Landing Zone
Unstructured
Data
Connectors
API
JDBC
SDK
Unstructured
Data
Change Data Capture
Unstructured
Data
Data Profiler, Rule Engine
Unstructured
Data
Rules Repo,
Workbench Repo
Postgresql
Unstructured
Data
Common Data Model
Unstructured
Data
Data Transformation
Pig
(Hortonworks Stack)
Impala
(Cloudera & MapR
Stack)
Unstructured
Data
Aggregated Entities
Impala
(Cloudera & MapR
Stack)
(Hortonworks Stack)
Unstructured
Data
Search
(Default)
Unstructured
Data
Web Service
RESTful
Unstructured Data
Security
Authentication
LDAP
Kerberos
SAML
Authorization
RBAC
ACL(Service Level, Directory
Level, Column Level)
Encryption
SSL
Transparent
Encryption
Encryption Zones
AES 256
Unstructured Data
Data Governance
(Hortonworks Stack)
(Cloudera Stack)
Technology Reference Architecture
(CDaaS)
28. 282828
Clinical
Data
Assets
IoT
Real World
Published
Literature
Social
Data
OMICs
Data
Clinical Data
Pipelines
Text Analytics
Pipelines
Foundational
Applications
Workflow Designer
Advanced Analytics
Data Resolver
Business Rules
Editor
Cohort Builder
Business Specific
Applications
Clinical Development
Clinical Development
Feasibility
Safety Signals
Business &
Science Driven
Actionable RBM
CRO Oversight
Study Monitoring
Safety/Efficacy
Optimized
Protocol Design and
Site/PI Selection
Automated Signal
Detection over Global
Data
Ad-Hoc Analytics
(e.g. drug differentiation,
biomarker CDx
Outcomes
Semantically
Organized
Information
Meta Data
Security
Governance
Glossary
Putting it together
29. 2929
Summary of discussions with Life Sciences leaders...
Disruptive thinking is required NOW
Clean
Longitudinal Data
Bench
to
Bedside
Patient Centricity Cognitive Systems ‘Silicon Valley’ Mind
Set
Operational Efficiencies