Explore the current state of PV automation in light of the challenges being faced by the industry and the PV automation landscape, as well as Covance's recent innovations in automation and the how these solutions have benefited our sponsors.
2. 2
State of PV Automation
The PV Journey so far…
Challenges Persist and New Challenges Emerge
► Complex and Divergent Regulations
► Reliance on Creation of ICSRs vs Analysis of Large Data Sources
► Varied Data Sources (Structured and Unstructured)
► Multiple, Disparate Tools and Systems
► Conservative Approaches to Change in Fully Adopting Automation
Paper to Electronic medium from
Collection process through
Submission
Bespoke Systems to COTS System for
Intake, Triage, Case Processing and
Submission
3. 3
What’s been possible to date?
►E2B exchange
• Business Partners and Agencies
• Accept as-is vs edit?
►Auto-narratives
• Generally ok for simple cases, less satisfactory for complex cases
• Clunky, requires extensive editing
►Auto-seriousness
• Using published ‘lists’ … which aren’t harmonized globally
• Questionable whether some terms would always be serious
4. 4
What’s been possible to date?
►Auto-expectedness
• Requires investment in maintenance
• Decision-making on MedDRA levels to use
̶ Conditional labeling presents a challenge
►Questionable benefit for low-volume products
►Automation of intake
• Varying degrees of success
• Often using E2B-based logic
• EDC gateways for clinical trial cases
• Integration with call intake systems, apps & websites
5. 5
The PV Automation Landscape
Basic
Process
Automation
Robotic
Process
Automation
Cognitive
Automation
Artificial
Intelligence
Basic Process Automation
Automates a workflow that involves multiple
roles
Continuous Metrics Collection
Reporting and Dashboards
Robotic Process Automation
Reduces or Eliminates a Manual Task
Cognitive Automation
Tools/Systems that:
Engage in Human Interaction
Leverage Natural Language
Processing Technologies
Often Combined with Robotic
Process Automation
Artificial Intelligence
Tools/Systems that:
No/Minimal Human Interaction
Self-learns through
experience
Next Slide: Safety Writing
6. 6
CPAT Overview
Source
Document
Review
Duplicate
Search
Data Entry
Self QC
100% 100%
50% 46%
0%
20%
40%
60%
80%
100%
120%
Initial Follow up
% Impact on Efforts
Pre Automation Efforts Post Automation Efforts
91.8% 91.6%
97.4%
96.1% 95.9%
100%
86.00%
88.00%
90.00%
92.00%
94.00%
96.00%
98.00%
100.00%
102.00%
Manual Hybrid Automated
Quality % Pre & Post CPAT
Pre CPAT Quality Post CPAT Quality
Next Slide: Back to Main Slide
Impact : 47% Effort Reduction
► Automatic identification of correct source documents & version
► Automated source document download and data extraction
► Consolidated comparative view of all source documents with highlighted
latest information
► Validation of source data and invalid data alerts
► Real time verification of parsed information
► Multi-level duplicate search automated
► Automatic download and data extraction from source
documents of potential duplicates
► Summary view of key fields for potential duplicates
► Automated data entry for all objective fields
► Automated partial data entry for hybrid fields
► Automated narrative generation
► Highlights Updated fields for self QC
► Consolidated view of all the updated fields and one view of
database fields, narrative and source document
7. 7
Closing Thoughts
►While the industry works to perfect true PV Artificial Intelligence in the
future, there are practical PV automations that can add significant value
today
►PV Automation is reducing Case Processing cost while also improving
Quality and Compliance
►Opportunities also exist to improve efficiencies in Aggregate Safety Report
production
8. 8
Problem Statement
Laborious task of identifying ICSR Cases from Literature Articles to initiation of CP
SciLit Pro
► Allows for end-to-end tracking of Literature monitoring workflow through an integrated system
► Provides REAL-TIME ONLINE DASHBOARD and is a SINGLE COMPREHENSIVE REPOSITORY of Literature
Abstracts
Basic Process Automation:
SciLitPro Literature Monitoring Solution
Back to Main Slide
How:
► Import of Literature Abstracts from
aggregator sites with rule-based
deduplication
► Auto-Triage for identification of
Abstracts of relevance
► Online Review of Literature Abstracts
with link to the original Article
► Initiation and Tracking of related
Actions for Article Procurement, and
Translation
9. 9
Robotic Process Automation
Case Processor Assist Tool (C-PAT)
Sciformix framework allows
for rapid development of
process-specific Tools like
‘C-PAT’ for Case Processor
activities, ‘I-PAT’ for Intake
activities and so on
CPAT incl Dup
Check Demo Video
Video Link - https://web.microsoftstream.com/video/9aa55b20-d30a-471b-b8a1-635afdeaad63
Problem Statement
► First Time Right Data Entry requires Removal/Reduction of Possible Human Error Touchpoints
C-PAT
► Automates Case Data Entry with help for structured source documents for Improved Quality &
Efficiency
► Platform for attaining ONE-TOUCH CASE PROCESSING with one round of manual interaction between Case
Intake and Case Submission
10. 10
Cognitive Automation
Medical Assist Tool
Ibuprofen-induced renal tubular acidosis (RTA) with severe hypokalemia is an under-recognized and potentially life threatening
adverse effect. We describe a case of recurrent severe metabolic acidosis secondary to repeated ibuprofen use. Case Report: A 45-
year-old patient presented to hospital on multiple separate occasions for unexplained generalized weakness and acidemia. Each
presentation was significant for hyperchloremic, hypokalemic metabolic acidosis. During the latest admission, investigations
revealed a pH of 6.89, a bicarbonate of 5 mmol/L, a potassium level of 2.5 mmol/L, and a urine pH of 7.0 with a positive urine gap
after fluid resuscitation. The patient had no significant medical co-morbidities and reported only taking pantoprazole. Initial
toxicologic testing was negative for acetaminophen, salicylates, toxic alcohols, and other drugs of abuse. Subsequent laboratory
investigation with High Performance Liquid Chromatography (HPLC) detected ibuprofen in both serum and urine. Upon repeat questioning
the patient endorsed a habit of taking ibuprofen at home. During each hospital admission, ibuprofen was never prescribed and all
symptoms and metabolic abnormalities resolved with supportive treatment. Case Discussion: Ibuprofen is a nonsteroidal anti-
inflammatory drug (NSAID) available over the counter (OTC) since 1983. It is commonly used as an anti-pyretic and analgesic.
Metabolic acidosis secondary to ibuprofen is an under-reported condition. The pathophysiology is unknown, but the most commonly
cited mechanism is an ibuprofen induced RTA. Alternative explanation is direct acidemia from ibuprofen and its metabolites, all of
which are weak acids. In all reported cases, metabolic abnormalities resolve with drug cessation and supportive treatment.
Conclusion: We describe a patient with recurrent episodes of weakness, acidemia and investigations consistent with distal RTA
secondary to ibuprofen. Abnormalities resolved with drug cessation and supportive treatment. Once ibuprofen was identified as the
culprit, the patient remained off this drug and has been out of hospital since. This case highlights a severe, yet reversible,
complication of a common OTC medication.
Problem Statement
► Correct Identification of Drug, Event, Medical History foundation of PV Case Processing workflow
Solution:
► Applies Natural Language Processing (NLP) to help identify Drug-AE pairs
► Leverages existing medical terminology & classifications such as MedDRA, WHODrug & ICD-10
► Informs Medical/HCP Review of Reporter Verbatim, Case Narrative, Literature Abstract to reduce Medical
expert burden