© 2014 PerkinElmer HUMAN HEALTH • ENVIRONMENTAL HEALTH Innovations in Clinical Analytics Accelerating Insights & Data-Driven Decisions Mark Demesmaeker
© 2014 PerkinElmer
HUMAN HEALTH • ENVIRONMENTAL HEALTH
Innovations in Clinical Analytics Accelerating Insights & Data-Driven Decisions
Mark Demesmaeker
Contents
Introduction
• Innovating to accelerate insights: Workflow, Advanced Analytics, Data Provisioning
Innovation Spotlight: CDR
•Guided workflow, SDTM standardization
Innovation Spotlight: RBM
•Closed-loop integrated monitoring workflow, leveraging legacy data
Innovation Roadmap: Clinical Application Framework
•Single framework, modular solutions
Safe Harbor Statement
• This document shows current intentions regarding product features, behavior, schedules and support; these intentions may change without notice as we respond to customer requirements.
• Any unreleased products, services or features referenced herein are not currently available and may not be delivered on time or at all. Customers who purchase PerkinElmer informatics applications should make their purchase decisions based upon products, features, and services that are currently available.
Introduction Accelerate Insights with PerkinElmer Innovations in Clinical Analytics
Clinical/Medical Data Review
Safety and PV Risk-Based Monitoring
Clinical Operations
Spotfire in Clinical Development
Health Outcomes & RWE
The Workflow Challenge: Where is the Actionable Data?
• How do find the analysis I need? • How do I know the data is current and actionable? • How do I navigate the analysis to find the data points I’m looking for? • How do I act on the data?
The Analytics Challenge: The Need for Faster, Smarter, Data-Driven Decisions Competitive pressure for: • Smarter decisions: pervasive advanced analytics beyond the data science community • Faster decisions: both human and automated
Need for agile analytics: • Data questions and answers change rapidly • Analytics must be adaptive
The Data Challenge: There’s More to Data Insights than Analytics
• Clinical Analytics
• Can studies be set up automatically?
• Are the data sets compatible?
• Can they be standardized?
• Can they be integrated?
• Can I look across historical snapshots of data?
• Can I perform analysis across multiple studies?
• How quickly can I access the data?
• …
Clinical Innovation Focus Areas
Persona-Driven Guided Workflows Integrated role-based insights to reduce drug development duration Curated analytics experience to streamline root-cause analysis Immediate access to actionable data for faster data-driven decisions
Advanced Analytics TERR for advanced statistics and predictive modeling Legacy data analysis to unlock value of legacy data assets Alignment with TIBCO Roadmap for fastest analytics innovation
Holistic Data Provisioning Ability to normalize and standardize clinical data Manage data extracts Easily provision data for analytics from multiple data sources
Clinical and Medical Data Review Risk-Based Monitoring
Innovation Spotlight: CDR & RBM
Persona-Driven Guided Workflows
Alerts Workflow Exploratory Workflow Line Listing Review
CRA Dashboard Central Monitor View Closed-Loop Monitoring
Holistic Data Provisioning
Study setup automation Extracts Management and Versioning SDTM Data Standardization
Study setup automation Data Source Identification Data Normalization
Advanced Analytics
Predictive risk modeling Adaptive monitoring Legacy data analysis and tolerance limits
Signal detection Time to event analysis Risk difference
Clinical Data Review Alerts Workflow Option, Extract Management, Data Standardization
Clinical Data Review
Provides access to a 360 degree view into Patient Demographics, Adverse Events, Labs, Vitals/ECG results and other safety domains
Increases agility by allowing interactive exploration of unexpected trends & hidden relationships
Enables rapid medical review of clinical data through intuitive navigation of population and subject-level data
Provides fast and accurate data quality assessment through fraud detection and visibility into data quality issues
Facilitates process improvement by allowing early identification of study issues and dimension-free data exploration
Navigate population-level and domain-level views
Select sub-populations for analysis
Easily navigate to subject-level data with Patient Profile
Workflow Configuration: Alerts-Driven and Exploratory Workflows
Explore trends in population-level data to select a sub-population for analysis
…or follow configurable therapeutic-area “Alerts” directly to the Patient Profiles of relevant subjects
Reduce drug development duration
Streamline root-cause analysis Save costs by identifying issues early
View all alerts or filter based on status or timeline
Line Listing Review and Queries
Mark records as “read” or “reviewed”
Raise queries about specific data points
Access full review history and audit trail
Mark records directly in analysis
for accelerated data-driven decisions
Access review history audit trail
for 21 CFR part 11 compliance and meet validation requirements
Automate Study Setup, Manage Data Extracts, Expedite Data Standardization
Reduce cycle time and medical
monitoring costs with study setup automation
Enable CRO oversight and leverage
historical data with historical analysis across data “snapshots”
Reduce cycle time to analysis with
expedited SDTM standardization
Merging and versioning of study data extracts to enable historical analysis
SDTM data standardization for study pooling and cross-study analysis
Study setup automation
PerkinElmer Data Services
PerkinElmer SignalsTM
Perspectives
PerkinElmer SignalsTM
Perspectives
Risk-Based Monitoring Closed-loop Monitoring Workflow, Adaptive Monitoring, and Data Normalization
Why Risk-Based Monitoring
• Traditional 100% SDV-based, scheduled monitoring approaches are costly and not effective in identifying issues with critical data and processes ◦ Mostly transcription errors are discovered ◦ Resources are not directed where they are most needed
• Risk-based monitoring shifts the focus from scheduled site visits to a targeted and adaptive monitoring and SDV approach, increasing patient safety and data quality while decreasing unnecessary resource costs
• RBM is recommended by regulatory bodies and the focus on RBM will increase with the upcoming GCP guidance, ICH E6 R2
• An effective RBM strategy should include a flexible, data-agnostic visualization platform, an integrated workflow solution, and advanced analytics to leverage the value of historical data assets in enhancing the monitoring strategy and trial design
…Members of TransCelerate have identified clinical
study execution as the initiative’s initial area of
focus. Five projects have been selected by the
group for funding and development, including:
development of a shared user interface for
investigator site portals, mutual recognition of study
site qualification and training, development of risk-
based site monitoring approach and standards,
development of clinical data standards, and
establishment of a comparator drug supply model.
RBM Overview
Risk Assessment
Categorization Tool (RACT)
ICH E6 R2- coming Nov 2016/early 2017
• Increased focus on Sponsor responsibilities for vendor oversight
• First mention of risk-based approach to quality in a GxP guidance (-> centralized Quality Risk Management (QRM) system)
• Focus on traceability in decision making and a metrics-based approach
• Need for tolerance limits for specific risk areas- what is the tolerated, systemic level of risk for a particular risk area
Risk-Based Monitoring
Simplifies Risk-Based Monitoring with intuitive dashboard displays
Enables flexibility by providing a vendor and data-agnostic solution designed to work with multiple clinical data sources
Supports adaptive design and adaptive monitoring through modeling of KRI algorithms, thresholds, weightings, and recommended actions
Provides risk model assessment and feedback loop through timeline analysis, geographic views, and regression modeling
Leverages historical data assets through the analysis of legacy data to establish tolerance limits for risk categories
Dashboard view into TransCelerate-aligned Risk Indicators, Category-Level Indicators, and Overall Site Indicators
Timeline view to assess the impact of recommended actions on risk signals
Geographic view of indicators and risk levels for resource management and root-cause assessment
Persona-Driven Views Simplify RBM through role-based guided workflows, users only see what they need
Reduce monitoring cycle time through fastest access to actionable data
The CRA is able to view and investigate risk Indicators for assigned sites
The Central Monitor is able to navigate the overall risk model and drill down to site-level details
Closed-Loop Monitoring Save time by following up on recommended actions directly in the analysis Use the feedback loop to learn what is working and easily adjust recommended actions
Optimize trial design and QbD by leveraging historical data on recommended actions
CRA updates status and comments for recommended actions
Status is automatically updated
Action status can be reviewed and actions modeled according to impact assessment
Advanced Analytics: Unlock the Value of Historical Data Assets
Regression modeling and advanced statistics offer a feedback loop on the historical performance of the risk model
Adjust KRI algorithms, weightings, thresholds, and recommended actions based on risk model assessment
Apply out-of-the box regression analysis, advanced statistics, historical analysis and predictive modeling to: Apply adaptive monitoring,
adjusting risk model in response to data
Derive statistically meaningful tolerance limits for risk categories (an ICH E6 R2 requirement)
Assess site feasibility Improve trial design
The Adaptive Risk Model: Beyond Visualization
Risk Trigger
Recommended Action
Action Follow-up
Action Impact Assessment
Risk Model Evaluation
Risk Model Adjustment
Site 104 is in the red category for the Protocol Deviations
KRI because while the count of minor violations is below
the average, there is a high proportion of major
violations.
A Site Visit has been recommended for the CRA based
on the Protocol Deviation Indicator. The CRA completes
the Action and adds Comments.
The Central Monitor is notified that the
Recommended Action has been
completed.
The Central Monitor can assess whether the
Action had an impact on the Protocol
Deviation Indicator.
The Central Monitor evaluates the Risk Model.
What is the historical correlation between:
o Visits at Site 104 and the Protocol
Deviation Indicator?
o Between Visits and other Key Risk
Indicators?
o Between Visits and the Protocol Deviation
Indicator across Sites?
Regression modeling and predictions based
on:
o Analysis of Indicators and other Indicators
o Analysis of Indicators and Actions
o Analysis of Indicators and Monitor
Comments
o Analysis against baseline and legacy
data
The Central Monitor adjusts Indicator Thresholds, KRI
Weightings, and Recommended Actions. The Risk
Model evolves based on monitoring impact and
adapts based on changing risk.
Clinical Application Framework Looking Ahead: Combining Workflow, Holistic Data Provisioning, and Advanced Analytics
PerkinElmer Clinical Application Framework: “One Stop Shop” for Clinical Data insights
Key Framework Features
Persona-Driven Guided Workflows • Integrated role-based insights to reduce drug development duration • Curated analytics experience to streamline root-cause analysis • Immediate access to actionable data for faster data-driven decisions
Advanced Analytics • TERR for advanced statistics and predictive modeling • Legacy data analysis to unlock value of legacy data assets • Alignment with TIBCO Roadmap for fastest analytics innovation
Holistic Data Provisioning • Ability to normalize and standardize clinical data • Manage data extracts • Easily provision data for analytics from multiple data sources
Secure Cloud Infrastructure • Multi-tenant and virtual private cloud options for secure, scalable architecture
Unified Framework • Single web application, modular clinical solutions to accelerate adoption
Streamlined Workflow, Curated Analytics
With PerkinElmer’s new Clinical Application Framework: Convenience: single, hosted web application for all clinical analytics modules Process Efficiency: intuitive analytics workflows tailored to user role Cost Savings: agile and automatic data normalization and provisioning Smart Insights: built-in advanced and predictive analytics Faster Decisions: direct access to actionable data, closed-loop workflow Quality by Design: leverage data to enhance trial design, site selection, analytics workflows Security and Compliance: secure cloud infrastructure, 21 CFR part 11 compliance
Persona-Driven Workflow
Automatic study setup and data normalization
TIBCO Spotfire visualizations, enhanced with advanced analytics
Actionable Data Insights
Advanced Analytics: Enhance Clinical Data Review with Advanced Analytics Built in R
TIBCO Enterprise Runtime for R (TERR) allows you to code open-source R directly in Spotfire with many built-in analytics methods and supported R packages
Embed advanced analytics directly into the analysis to extend advanced data insights beyond the data science community
Smarter data insights with advanced analytics, including: Relative Risk Time to Response Analysis Survival Analysis Time to AE Analysis
Holistic Data Provisioning
This document shows current intentions regarding product features, behavior, schedules and support;
these intentions may change without notice as we respond to customer requirements.
Analytics Workflow: Before
This document shows current intentions regarding product features, behavior, schedules and support;
these intentions may change without notice as we respond to customer requirements.
Separate analysis file for every study and analytics use case
Retention of audit trail requires custom functionality User administration and security
management handled in separate interface
Data mapping is configured on a per-study and per-analysis basis
Analytics Workflow: PerkinElmer Clinical Application Framework
Single web application for all clinical analytics solution modules
This document shows current intentions regarding product features, behavior, schedules and support;
these intentions may change without notice as we respond to customer requirements.
Full audit trail and centralized system administration
Single interface for user management and definition of system roles
Analytics Workflow: Before
This document shows current intentions regarding product features, behavior, schedules and support;
these intentions may change without notice as we respond to customer requirements.
Ability to mark records or raise queries requires custom functionality
Workflow based on tab navigation in Spotfire User needs to know how to filter and mark data
Multiple visualizations and analyses can be difficult to navigate
Analytics Workflow: PerkinElmer Clinical Application Framework
This document shows current intentions regarding product features, behavior, schedules and support;
these intentions may change without notice as we respond to customer requirements.
Full power and flexibility of Spotfire analytics
Guided analytics experience based on user roles and permissions
Persona-driven views and workflows
Out-of-the-box ability to mark records, raise queries, and send notifications
Simplified, intuitive filtering and curated visualizations
Automatic Study Provisioning and Mapping: PerkinElmer Clinical Application Framework
This document shows current intentions regarding product features, behavior, schedules and support;
these intentions may change without notice as we respond to customer requirements.
Automatic study setup and data provisioning
Easily configure data upload frequency
Adjust field mapping to analysis standard
Clinical Application Framework Conceptual Roadmap
This document shows current intentions regarding product features, behavior, schedules and support;
these intentions may change without notice as we respond to customer requirements.
1.0 2.0 1.5
Medical Monitor Workflow Workflow Configuration
Patient Narratives
Med
ical
Rev
iew
R
BM
Data Manager Workflow Biostatistics Workflow
RW
E Clin
ical
Wo
rkfl
ow
Mo
du
les
Automatic Study Setup and Data Provisioning
Advanced Analytics and Statistical Modeling
Audit Trail Cloud infrastructure
Audit Trail CRA Dashboard Threshold Modeler Recommended Actions
Action Logging Tolerance Limit Definition Cohort Builder
Data Source Discovery
Thank You! [email protected]