What AI can do to
transform our business
1 Apr 2021
Dr Bonnie Cheuk
Senior Director, Business & Digital Transformation Leader
Presentation at the Digital Leadership Forum
1 Apr 2021
2
Hi I’m Bonnie Cheuk
• I build future-ready capabilities for
76,000+ employee in AstraZeneca
• I focus on building digital capabilities AND
learning agility (i.e. making continuous
learning happen naturally at work)
Business & Digital Transformation Leader
AstraZeneca
Healthcare in a Changing World
1 Apr 2021
3
Artificial Intelligence vs Automation
What AI can do
Q&A
1
2
3
4
Healthcare in a changing world
1 Apr 2021
4
Global economic
shock
Growing & ageing
populations
Increasing burden
of chronic disease
Impact of
COVID-19
Digital & technical
breakthroughs
Global GDP forecast
5.3% below
pre-pandemic
projections1
Many patients
in US, EU and Asia
deferred or cancelled
scheduled treatment
early in the pandemic4
2.1 billion people
will be aged 60+
by 2050
Digital health market to
increase nearly
six times by 20263
42 million people
die from NCDs
each year2
1 Apr 2021
5
Putting patients at the heart of what we do, we continue to transform
our science to create the greatest and swiftest impact on disease and
deliver life-changing medicines
• Cellular and animal models
• Molecular imaging and AI
• Quantitative pharmacology
• Biomarker identification
Better
predicting
clinical
success
Pioneering
new
approaches in
the clinic
Growth through innovation investments
Data Science and AI
• Dynamic multi-omics
• Knowledge graphs
• Epigenetics
• Cell therapy
• PROTACs
• Augmented drug design
• Digital R&D
• Digital technologies
• Digital solutions
Enhancing
our
understanding
of disease
Designing
next
generation
therapeutics
March 2021
6
Automation
• Explicit rules/processes
• Accurate output
• Digitalise the process
• Do tasks better/faster
“Dumb” technology
powered by
programmable bots
• Follows rules to handle
straightforward tasks.
• Can't react to new
situations.
Artificial Intelligence
• No need to spell out
exactly how it works
• Give machine a set of
data (as input)
• Give machine a set of
results (as output)
Machine figures out what
goes on in between
• Can be software, a
robot, an algorithm.
What is AI vs Automation?
1 Apr 2021
7
What is Digital | Digital Strategy | Digital leadership | Digital Lexicon | Organisation Responsiveness |
Change Readiness | Data | Data Science & AI | Agile | Design Thinking | Lean | Digital Lexicon
Build Future-Ready capabilities for all employees
1 Apr 2021
8
After two weeks, the testers were able to successfully identify cancerous tissue with 85%
accuracy, which increased to 99% in the top performing group.
Diagnosing Breast Cancer
1 Apr 2021
9
Let’s challenge our scientists’ belief: Only human can do certain task?!
Diagnosing Breast Cancer
1 Apr 2021
10
“AI will not replace drug hunters, but drug hunters
who don’t use AI will be replaced by those who do.”
― Andrew Hopkins, CEO Exscientia
What AI
can do…
• Gain a better understanding of the diseases we
want to treat
• Identify new targets for novel medicines
• Speed up the way we design, develop and make
new drugs
• Recruit for and designing better clinical trials
• Drive personalised medicine strategies
1 Apr 2021
11
Accelerate innovative science
In R&D, we are harnessing data and AI to discover and
develop new medicines faster, for the right patients
12
4-year old Learning
Show 2 or 3 pictures of a bicycle to a child.
Now ask the child to find all the other
bicycles in this pile of pictures.
He picks up unicycles, tricycles and even
bicycles that are wrapped around trees.
Based on very limited info, the child is able
to generalise much more broadly than it
would seem possible.
13
AI/Machine Learning
Give a million pictures of bicycle to a
computer.
Teach the computer by pointing to each
picture and specifying which is a bicycle and
which is not.
After a while, the computer learns which
pictures to label as a bicycle.
4-year old Learning
Show 2 or 3 pictures of a bicycle to a child.
Now ask the child to find all the other
bicycles in this pile of pictures.
He picks up unicycles, tricycles and even
bicycles that are wrapped around trees.
Based on very limited info, the child is able
to generalise much more broadly than it
would seem possible.
1 Apr 2021
14
The deep learning-based approach was able to correctly classify ‘damaged’ and ‘non-damaged’
tablets 95.2% of the time.
Case Study 1: Meeting regulatory requirements:
Description Analysis of drug product
1 Apr 2021
15
We created a machine learning model to describe doctors who ‘have’ and ‘have not’ prescribed Lokelma.
We can find doctors that are likely to prescribe Lokelma but have not done so yet. These critical insights
inform the field sales representatives to prioritise limited resources towards engagement with these doctors.
Case Study 2: Who is likely to prescribe a new drug
to patients?
1 Apr 2021
16
Use NLP to find around twenty potentially highly relevant patents - Legal Attorney team now have the
capability to filter and review the most relevant patents for AstraZeneca’s portfolio and take early action.
Case Study 3: Review of all patents to find the relevant documents
that might limit AstraZeneca’s freedom to market products
What AI
can do…
• Clinical Supply Chain: Just in time delivery of
products to the patients
• Commercial: Understand the effectiveness of
digital promotional channels with the HCPs –
inform budget allocation and future planning
• HR: Identify and match the best internal
candidates to open internal roles – find the
most suitable offer for the person and for AZ
• Lab: Predict when equipment is going to fail
and optimise maintenance schedules to
prevent failure
1 Apr 2021
17
Accelerate innovative science
What is the business or scientific problem you are
trying to solve?
AI is not magic - it is logic and science
that we can choose to apply
1 Apr 2021
18
Do you have clean, unbiased data?
Are you prepared to work with the answers
(predictions) that you have?
Are you clear about the assumptions you have
made?
Are you clear about the implications of how you
use the AI-driven solutions (predictions)?
1
2
3
4
5
Future of work: Humans and machines work together to
create value in a data-driven AI world
• AI can help or hurt: humans
can make choices about
which way to go
• Human judgement, values
and ethics matter
• Learning Agility is critical:
proactively anticipate and
handle unintended
consequences
Future of work: Humans and machines work together to
create value in a data-driven AI world
• AI can help or hurt: humans
can make choices about
which way to go
• Human judgement, values
and ethics matter
• Learning Agility is critical:
proactively anticipate and
handle unintended
consequences
Apply 5 work habits to keep learning
Self/Team
Reflection
Innovation &
Growth
Mindset
Psychological
Safety
Inclusive
Meetings &
Collaboration
Learning &
Working as
Networks
21
Business and Digital Transformation Leader
Digital Capability & Learning Agility, AstraZeneca
Credit to Dr Brenda Dervin for her
ongoing inspiration and guidance
@bonniecheuk
/bonniecheuk http://www.sense-making.org
Let’s learn together!
22
Confidentiality Notice
This file is private and may contain confidential and proprietary information. If you have received this file in error, please notify us and remove
it from your system and note that you must not copy, distribute or take any action in reliance on it. Any unauthorized use or disclosure of the
contents of this file is not permitted and may be unlawful. AstraZeneca PLC, 1 Francis Crick Avenue, Cambridge Biomedical Campus,
Cambridge, CB2 0AA, UK, T: +44(0)203 749 5000, www.astrazeneca.com
23

AI today and its power to transform healthcare

  • 1.
    What AI cando to transform our business 1 Apr 2021 Dr Bonnie Cheuk Senior Director, Business & Digital Transformation Leader Presentation at the Digital Leadership Forum
  • 2.
    1 Apr 2021 2 HiI’m Bonnie Cheuk • I build future-ready capabilities for 76,000+ employee in AstraZeneca • I focus on building digital capabilities AND learning agility (i.e. making continuous learning happen naturally at work) Business & Digital Transformation Leader AstraZeneca
  • 3.
    Healthcare in aChanging World 1 Apr 2021 3 Artificial Intelligence vs Automation What AI can do Q&A 1 2 3 4
  • 4.
    Healthcare in achanging world 1 Apr 2021 4 Global economic shock Growing & ageing populations Increasing burden of chronic disease Impact of COVID-19 Digital & technical breakthroughs Global GDP forecast 5.3% below pre-pandemic projections1 Many patients in US, EU and Asia deferred or cancelled scheduled treatment early in the pandemic4 2.1 billion people will be aged 60+ by 2050 Digital health market to increase nearly six times by 20263 42 million people die from NCDs each year2
  • 5.
    1 Apr 2021 5 Puttingpatients at the heart of what we do, we continue to transform our science to create the greatest and swiftest impact on disease and deliver life-changing medicines • Cellular and animal models • Molecular imaging and AI • Quantitative pharmacology • Biomarker identification Better predicting clinical success Pioneering new approaches in the clinic Growth through innovation investments Data Science and AI • Dynamic multi-omics • Knowledge graphs • Epigenetics • Cell therapy • PROTACs • Augmented drug design • Digital R&D • Digital technologies • Digital solutions Enhancing our understanding of disease Designing next generation therapeutics
  • 6.
    March 2021 6 Automation • Explicitrules/processes • Accurate output • Digitalise the process • Do tasks better/faster “Dumb” technology powered by programmable bots • Follows rules to handle straightforward tasks. • Can't react to new situations. Artificial Intelligence • No need to spell out exactly how it works • Give machine a set of data (as input) • Give machine a set of results (as output) Machine figures out what goes on in between • Can be software, a robot, an algorithm. What is AI vs Automation?
  • 7.
    1 Apr 2021 7 Whatis Digital | Digital Strategy | Digital leadership | Digital Lexicon | Organisation Responsiveness | Change Readiness | Data | Data Science & AI | Agile | Design Thinking | Lean | Digital Lexicon Build Future-Ready capabilities for all employees
  • 8.
    1 Apr 2021 8 Aftertwo weeks, the testers were able to successfully identify cancerous tissue with 85% accuracy, which increased to 99% in the top performing group. Diagnosing Breast Cancer
  • 9.
    1 Apr 2021 9 Let’schallenge our scientists’ belief: Only human can do certain task?! Diagnosing Breast Cancer
  • 10.
    1 Apr 2021 10 “AIwill not replace drug hunters, but drug hunters who don’t use AI will be replaced by those who do.” ― Andrew Hopkins, CEO Exscientia
  • 11.
    What AI can do… •Gain a better understanding of the diseases we want to treat • Identify new targets for novel medicines • Speed up the way we design, develop and make new drugs • Recruit for and designing better clinical trials • Drive personalised medicine strategies 1 Apr 2021 11 Accelerate innovative science In R&D, we are harnessing data and AI to discover and develop new medicines faster, for the right patients
  • 12.
    12 4-year old Learning Show2 or 3 pictures of a bicycle to a child. Now ask the child to find all the other bicycles in this pile of pictures. He picks up unicycles, tricycles and even bicycles that are wrapped around trees. Based on very limited info, the child is able to generalise much more broadly than it would seem possible.
  • 13.
    13 AI/Machine Learning Give amillion pictures of bicycle to a computer. Teach the computer by pointing to each picture and specifying which is a bicycle and which is not. After a while, the computer learns which pictures to label as a bicycle. 4-year old Learning Show 2 or 3 pictures of a bicycle to a child. Now ask the child to find all the other bicycles in this pile of pictures. He picks up unicycles, tricycles and even bicycles that are wrapped around trees. Based on very limited info, the child is able to generalise much more broadly than it would seem possible.
  • 14.
    1 Apr 2021 14 Thedeep learning-based approach was able to correctly classify ‘damaged’ and ‘non-damaged’ tablets 95.2% of the time. Case Study 1: Meeting regulatory requirements: Description Analysis of drug product
  • 15.
    1 Apr 2021 15 Wecreated a machine learning model to describe doctors who ‘have’ and ‘have not’ prescribed Lokelma. We can find doctors that are likely to prescribe Lokelma but have not done so yet. These critical insights inform the field sales representatives to prioritise limited resources towards engagement with these doctors. Case Study 2: Who is likely to prescribe a new drug to patients?
  • 16.
    1 Apr 2021 16 UseNLP to find around twenty potentially highly relevant patents - Legal Attorney team now have the capability to filter and review the most relevant patents for AstraZeneca’s portfolio and take early action. Case Study 3: Review of all patents to find the relevant documents that might limit AstraZeneca’s freedom to market products
  • 17.
    What AI can do… •Clinical Supply Chain: Just in time delivery of products to the patients • Commercial: Understand the effectiveness of digital promotional channels with the HCPs – inform budget allocation and future planning • HR: Identify and match the best internal candidates to open internal roles – find the most suitable offer for the person and for AZ • Lab: Predict when equipment is going to fail and optimise maintenance schedules to prevent failure 1 Apr 2021 17 Accelerate innovative science
  • 18.
    What is thebusiness or scientific problem you are trying to solve? AI is not magic - it is logic and science that we can choose to apply 1 Apr 2021 18 Do you have clean, unbiased data? Are you prepared to work with the answers (predictions) that you have? Are you clear about the assumptions you have made? Are you clear about the implications of how you use the AI-driven solutions (predictions)? 1 2 3 4 5
  • 19.
    Future of work:Humans and machines work together to create value in a data-driven AI world • AI can help or hurt: humans can make choices about which way to go • Human judgement, values and ethics matter • Learning Agility is critical: proactively anticipate and handle unintended consequences
  • 20.
    Future of work:Humans and machines work together to create value in a data-driven AI world • AI can help or hurt: humans can make choices about which way to go • Human judgement, values and ethics matter • Learning Agility is critical: proactively anticipate and handle unintended consequences Apply 5 work habits to keep learning Self/Team Reflection Innovation & Growth Mindset Psychological Safety Inclusive Meetings & Collaboration Learning & Working as Networks
  • 21.
    21 Business and DigitalTransformation Leader Digital Capability & Learning Agility, AstraZeneca Credit to Dr Brenda Dervin for her ongoing inspiration and guidance @bonniecheuk /bonniecheuk http://www.sense-making.org Let’s learn together!
  • 22.
  • 23.
    Confidentiality Notice This fileis private and may contain confidential and proprietary information. If you have received this file in error, please notify us and remove it from your system and note that you must not copy, distribute or take any action in reliance on it. Any unauthorized use or disclosure of the contents of this file is not permitted and may be unlawful. AstraZeneca PLC, 1 Francis Crick Avenue, Cambridge Biomedical Campus, Cambridge, CB2 0AA, UK, T: +44(0)203 749 5000, www.astrazeneca.com 23