Financial Tata Consultancy Services Limited ended. Statements
Cognitive Automation 090718 - Tata Consultancy Services · WHITE PAPER The Road Ahead We believe...
Transcript of Cognitive Automation 090718 - Tata Consultancy Services · WHITE PAPER The Road Ahead We believe...
Leveraging Meta Data Management: Powering Cognitive Automation in Clinical Trial Processes
WHITE PAPER
Abstract
Today, articial intelligence (AI) is transforming
every area of business, including life sciences,
particularly in the clinical trials space. Numerous
companies are already well on their way to using
cognitive automation for clinical trial set up, fraud
detection, pharmacovigilance, recommendation
engines, crop classication, and so on. Systems
such as natural language processing (NLP),
machine learning (ML), and neural networks are
becoming commercially viable solutions for
deciphering clinical content and simplifying
reporting for supporting business decisions and
ensuring compliance.
As a welcome consequence to all the clinical
standard initiatives which were mainly targeting
standards governance, this paper highlights how
these initiatives in combination with AI can
provide a real shot at optimizing clinical
development processes. We explore how
metadata management can play a foundational
role in standardizing governance, while being a
key enabler for cognitive process automation.
The Evolving Compliance Landscape
The pharmaceutical business is a highly process-driven space,
governed by stringent local and international regulations
pertaining to drug discovery, development, and testing.
Companies therefore are seeking ways to increase trial
efciency and improve outcomes while reducing overall cycle
time. However, the diverse nature of trials across therapeutic
areas is a primary obstacle towards optimizing the company-
wide process.
By early last year, there were over 200,000 registered clinical 1
research studies. Depending on the therapy area and the way
the organization is structured, there are many variations in
terms of governance frameworks and protocols followed even
within an organization. Without a standard technology-driven
process for setting up clinical trials, these studies will run the
risk of costly delays or fail to comply with regulatory guidelines
for submissions.
Over the last decade or so, the industry has been launching
various standardizations such as the Clinical Data Interchange
Standards Consortium (CDISC), and simplication such as the
Transcelerate initiatives. These are aimed towards harmonizing
and standardizing electronic data submissions.
Drive for Cognitive Process Automation
We have seen companies taking steps towards applying AI
based solution across the entire scope of clinical activities. 2GlaxoSmithKline, for example, uses AI for data management -
mapping raw clinical information per study data tabulation
(STDM) standards. The company met initial success with 50-
60% accuracy and has been steadily improving.
3 Proprietary technologies such as those from Innoplexus
condenses critical life sciences information to create a
repository of business use cases, which make information more
actionable for the end-user.
WHITE PAPER
WHITE PAPER
The fundamental components of such cognitive systems –
natural language processing (NLP), machine learning (ML), and
neural networks – are becoming increasingly accessible to drug
manufacturers. These can be used to analyse clinical data for
not only supporting business decisions but also ensuring
compliance. For example, building protocol components can be
simplied by implementing NLP. In turn, this will subsequently
help create other deliverables such as electronic case report le
(eCRF) or analysis plan while ML creates the metadata used for
generating submission packages.
Certain factors, however, need to be considered for effectively
implementing cognitive automation in these areas (Figure 1).
These include:
n Diverse standards across clinical lifecycle and
therapeutic areas
During the course of a clinical trial, multiple study artefacts
covered by the electronic common technical document (eCTD)
guidelines such as study protocol, submission dataset, and
clinical study reports (CSRs) need to be drafted and prepared
for submission to ethics committees and review boards. Each
submission deliverable follows different standards – from
simple templates for protocols to highly structured SDTMs or
analysis data model (ADaM) for ling ndings reports. These
can also be further categorized by different attributes of the
study. From the study sponsors perspective, it is difcult to
have a thorough understanding of each standard before
Figure 1: Scope for automation across clinical trial lifecycle
Activity
Sta
ndard
sD
elivera
ble
sD
ow
n s
tream
Impact
Study Design Study Setup Study Execution
Protocol Design CRF Design SAP / DPS SDTM/ADaM/TFLAnalysis/Report
Generation
PRM, CPT CDASH, ODMNo ExistingStandard SDTM, ADaM eCTD
CRF formsAnalysis/Shells
SDTM MetadataADaM
MetadataTFL Setup
Submission Package
Draft ProtocolDatabase buildmetadata form
to upload in eDC Draft SAP/DPS
SubmissionDatasets, Reports
Setup
Submission Package(Datasets, Reports
CSR)
concluding which of the standards is applicable for the study.
In case changes made by regulators to the existing guidelines
are overlooked and not incorporated into studies wherever
applicable, it can lead to massive rework and delay trial report
submission.
n Lineage and traceability
Disjointed reporting standards across multiple sources and the
lack of traceability or lineage for the end to end process also
stand in the way of creating a seamless process ow.
Standards for protocol are driven by the protocol
representation model (PRM) within which data collected for a
study is classied as electronic data capture (eDC) committed
by third party vendors. While eDC is governed by Clinical Data
Acquisition Standards Harmonization (CDASH) guidelines or by
internal standards, non-eDC data is tracked and maintained in
line with data transfer agreements between sponsor and third
party data vendors. Since these standards have to be stored
discretely and are not linked together, it becomes difcult to
synchronise them in case studies are amended at a later date.
Unless standard or guideline updates are synchronized with
downstream processes, they are at a risk of failing to comply
with established good practices (GxP).
Clinical Development Lifecycle 4.0
If we are to build a clinical development lifecycle that is highly
automated and is capable of rapidly adapting to regulatory and
internal standards, we will need to follow a phased approach
that comprises:
1. Standards governance through a metadata repository
AI in clinical trials can be fully utilized if it is built using the
metadata repository. For example, data from completed trials,
supervised learning techniques can be used to train an AI
system for generating mapping specication from eDC to
SDTM. Similarly, for creating protocols, NLP could be leveraged
to convert unstructured data into structured, searchable data –
improving overall quality.
n The rst step towards automation would be to ensure
standards compliance by semantically comparing source
metadata with enterprise standards – both of which are
being available in the metadata repository. This repository
will have library for each deliverable type, including CRF,
WHITE PAPER
WHITE PAPER
SDTM, ADaM, among others. This will also provide the basis
for understanding the impact that a change in one
deliverable will have on another.
n Standards would also need to be maintained across different
levels – spanning multiple therapeutic areas and lifecycle
stages per the Clinical Data Interchange Standards
Consortium (CDISC).
2. Cognitive process automation
Once the metadata registry is up and running, AI can be
leveraged not just for preparing deliverables, but also ensuring
that they are kept up to date. Such activities can include:
n Study setup – AI can use historic data and well dened
ontologies to create complete or partial deliverables such as
protocol sets from a topic library or electronic case report
forms (eCRFs) based on protocols.
n Study conduct and closure – As the study data starts
owing in near real-time, this enormous volume of
information needs to cleansed and transformed for
downstream analysis. AI-based unsupervised learning
methods that have been trained using semantic
representations of data along with corresponding contextual
information can be used to create a library of rules that can
be applied to incoming data. Such rules can be used for
governing evaluating and scrubbing data and can create
downstream artefacts like a mapping document to link
eCRFs with SDTMs.
n Study analysis and submission – This is a particularly
complex stage of a study and typically involves extensive
statistical analysis and medical regulatory content creation.
AI-based supervised learning methods can assist human
agents in this regard by running quality analysis on
narratives. This can help trial operators identify and acquire
missing information or even run analysis on mock shell table
listing gures (TLF) per enterprise standards for generating
and submitting reports to regulatory authorities – improving
turnaround time as well as quality.
WHITE PAPER
The Road Ahead
We believe that the key to improving and optimizing clinical
development process is to create a centralized AI-based data
management platform that feeds off an enterprise-wide
metadata management repository. The efciency of such a
solution can be further amplied by implementing AI to
optimize process interdependencies and externally dependent
risks like clinical trial recruitment.
References1. ClinicalTrials.gov, Trials, Charts, and Maps, accessed April 27, 2018,
https://clinicaltrials.gov/ct2/resources/trends#RegisteredStudiesOverTime
2. Forbes, Biting The Data Management Bullet At GlaxoSmithKline, January 8, 2018,
accessed April 27, 2018,
https://www.forbes.com/sites/tomdavenport/2018/01/08/biting-the-data-
management-bullet-at-glaxosmithkline/2/#3a6ccd0f7122
3. YourStory, How Innoplexus uses smart and on-edge technology to help its pharma
clients nd relevant answers hidden in unstructured data, April 10, 2018, accessed
April 30, 2018, https://yourstory.com/2018/04/innoplexus-uses-smart-edge-
technology/
WHITE PAPER
All content / information present here is the exclusive property of Tata Consultancy Services Limited (TCS). The content / information contained here is correct at the time of publishing. No material from here may be copied, modified, reproduced, republished, uploaded, transmitted, posted or distributed in any form without prior written permission from TCS. Unauthorized use of the content / information appearing here may violate copyright, trademark and other applicable laws, and could result in criminal or civil penalties. Copyright © 2018 Tata Consultancy Services Limited
About Tata Consultancy Services Ltd (TCS)
Tata Consultancy Services is an IT services, consulting and business solutions
organization that delivers real results to global business, ensuring a level of
certainty no other firm can match. TCS offers a consulting-led, integrated portfolio
of IT and IT-enabled, infrastructure, engineering and assurance services. This is TMdelivered through its unique Global Network Delivery Model , recognized as the
benchmark of excellence in software development. A part of the Tata Group,
India’s largest industrial conglomerate, TCS has a global footprint and is listed on
the National Stock Exchange and Bombay Stock Exchange in India.
For more information, visit us at www.tcs.com
TCS
Des
ign
Serv
ices
I M
I 07
I 18
About The Authors
Devraj Goulikar, Life Sciences
Platform Solutions Lead, TCS
Devraj is the Life Sciences
Platform Solutions Lead at Tata
Consultancy Services. He is
responsible for conceptualizing
and delivering platform-based
solutions for the Life Sciences
unit. He has been with TCS for
over 22 years and holds a
master’s degree in technology
from IIT Bombay.
Charusheela Thakur, Domain
Consultant, Life Sciences
Platform Solutions, TCS
Charusheela is a Domain
Consultant for the Life Sciences
Platform Solutions team and has
been with TCS for over five
years. She has been working in
the pharmaceutical industry for
16 years and specializes in areas
such as clinical trial reporting,
standardization, and analytics.
She has a master’s degree in
statistics from the University of
Mumbai.
Contact
Visit the page on Life Sciences & Healthcare www.tcs.com
Email: [email protected]
Subscribe to TCS White Papers
TCS.com RSS: http://www.tcs.com/rss_feeds/Pages/feed.aspx?f=w
Feedburner: http://feeds2.feedburner.com/tcswhitepapers