Presentation by Hari Radhakrishnan (senior solution developer) and Josh Mesout (graduate developer), in my team at Deep Learning Summit in London on September 23rd 2016. Brief overview about how we have been exploring artificial intelligence and how predictive modelling has the potential to revolutionise what we do across the drug discovery and development process. Examples include recent exploratory work on AI chatbots and video facial sentiment detection.
2. AZ CHATBOT
AstraZeneca
We are a global, science-led
biopharmaceutical business
pushing the boundaries of science
to deliver life-changing medicines.
61,500
employees worldwide
$24.7bn
2015 Revenue*
100+
Countries
3. AZ CHATBOT
Technology Incubation Lab, CTO
Hari works as an Solution
Architect in the UK Tech
Incubation Lab. He loves to bring
new emerging technologies into
the hands of users
Josh is a Developer in the UK
Technology Incubation Lab. He
develops prototypes and proof-
of-concepts with business
customers
4. “By 2020, 85% of customer
interactions will be managed
without a human” (Gartner)
AZ CHATBOT
AI is reshaping our world today…
AZ CHATBOT
5. Volumes of Data
Next Generation Sequencing
Whole body imaging
Tissue Microarrays
Sales force optimisation
Clinical trial statistical analytics
High Throughput Screening
Toxicogenomics
Open Innovation Approaches
PowerPoint/Excel content
Structured databases
Predictive Chemistry Modelling
HR employee retention
Real-time news sentiment
experimental data capture
Wearable sensor information
Log analytics in Operations
Streaming of Data Variety of Data Complexity of Data
We’re a data driven company: We need data driven decisions
AZ CHATBOT
What does this mean for AZ ?
AZ CHATBOT
7. Developed video extraction Text
AZ CHATBOT
PROACT
AZ CHATBOT
Developed a mobile app to capture
patient videos and diaries to
understand drug tolerability and
potential side effects in phase 1
clinical trials.
Developed facial expression Sentiment
8. Can we teach it information from the internet?
Can we get it returning information about AZ? How about information from our Intranet?
Can we apply it to a real use case?
Conversational UI will disrupt the landscape in AstraZeneca
AZ CHATBOT
This is a start for us
10. AZ CHATBOT
Use Cases
Help Desk
Enterprise Q&A
Patient Engagement
Drug Information
Social Media Interaction
Expert Lookup
INTERNALEXTERNAL
Adverse Events
Reporting
Finance helpdesk
11. A Bot to help users dig out
useful contacts within
AstraZeneca with specific
skillsets.
AZ CHATBOT
Expertise Lookup within AstraZeneca
Expert Lookup
13. AI uses Natural Language
Processing (NLP) to
understand everyday speech
AI Bots automate
everyday tasks
Cuts down on
unnecessary
manpower
User submits
ticket: Can’t
login
Bot identifies
issue of locked
account
Bot resets
users
password
AZ CHATBOT
Automating the Help Desk
Help Desk Bots
Undertake Basic Help desk tasks:
1. ChatBots for Service Now
2. Knowledgebase - FAQ
3. Password resets
4. Various form based applications
5. Ordering stuff
6. Handle repetitive tasks so human
resources can be put to better
use.
14. • ChatBot should understand the context of
the queries and provide information or
redirect to the right resources
• Support Multiple Languages
• ChatBot that understand medical and
scientific terminology
AZ CHATBOT
Increasing Patient Engagement
Patient Engagement
Social Media
interactions & Campaigns
Patient advocacy groups
Patient Portals
Mobile Apps
15. Question and
answer systems
A Question and Answer
Bot is built using structured
FAQ based content that
would try answering user
questions about
AstraZeneca based on the
context of users questions.
AZ CHATBOT
Q&A Systems – Pharma as a Bot
16. AZ CHATBOT
REACH OUT TO US
Questions
Thank You &
hariprasad.radhakrishnan@astrazeneca.com
joshua.mesout@astrazeneca.com
nick.brown@astrazeneca.com
AI Platforms Conversational UI Bot Aggregators ServiceNow Bots Biomedical UI