Curlew Research Brussels 2014 Electronic Data & Knowledge Management

Curlew Research 2014

Knowledge management with CROs & partners Nick Lynch Curlew Research


Summary

● Challenges to Collaboration and its growth in life science

● Models of Data Exchange with CROs and partners

● Curation – empowering scientists for collaboration – How R&D Search relies on good meta data – How Training is part of knowledge

management 2

3

AstraZeneca Outsourcing AZ’s outsourcing bill was about $3 billion per annum

• AstraZeneca is a global, innovation-driven, integrated biopharmaceutical company

• AZ employs over 50,000 people • 44% in Europe, 30% in the Americas, 22% in Asia and 4%

in ROW • Has over 9,000 people in our R&D organisation • Last year AZ invested $4 billion in R&D • In 2013, worldwide sales totalled $26 billion

About AstraZeneca

About AstraZeneca • Research within AZ comprises

• 6 innovative medicines units • Oncology • Infection • Cardiovascular and Metabolic • Respiratory and Inflammation • Asia & Emerging Markets • Neuroscience (virtual)

• Supported by innovative medicines functions eg. • Drug Safety and Metabolism • Discovery sciences

• Principally located on three main sites • Alderley Park, Cheshire (UK) → Cambridge (UK) • Mölndal, Gothenburg (Sweden) • Gatehouse Park, Waltham (US) • Plus: Shanghai

4

Drug Discovery and Early Dv within AstraZeneca


Quick Survey!

●Who is working with CROs & partners? ●Who has multiple CROs/partners? ●Who thinks they will have more partnerships in the future? ●Who has shared labs with partners?

5

6

Life Science Information Landscape

Big Life Science

Company

Yesterday Today Tomorrow

Yesterday Today Tomorrow Innovation Model

Innovation inside Searching for Innovation Heterogeneity of collaborations. Part of the wider ecosystem

IT Internal apps & data Struggling with change Security and Trust

Cloud/Services

Data Mostly inside In and Out Distributed

Portfolio Internally driven and owned Partially shared Shared portfolio

A rapidly evolving ecosystem

Why Externalise?

• Increase choice • Higher quality candidates

Increase project resource

• Dynamically resource projects according to need Flexibility

• Liberate internal scientists • Access external ideas

Innovation

• Ensure future agility Reduced fixed

costs

7

Understanding drivers for

externalisation is key to measuring

success & managing

information

7


Science information management has come along way.....


Where is your sweetspot? Spectrum of Engagement

Partners use Pharma

Software and data

Pharma use partners Software and data

Share Data via File

exchange

Share Data via B2B Services

Each relationship CRO/partner will be at a different capability Different models work in different situations 9


What Relationship do you have Full embedding is not always the best option

Getting Started

• Basic building blocks but scaling is hard

• Basic data sharing with CROs via email

• Manual effort to bring data in

• Overall coordination is manual

CRO engaged

• Capable of scaling to wider interactions

• Agreed Data contracts

• Transactional support

CRO integrated

• Efficient knowledge transfer

• Efficient data transfer

• Access to tools where necessary

• True b2b/supply chain relationship

• Scalability/agility

CRO embedded

• Using Pharma systems as if employees

• This could be too coupled together and hence not flexible for either party

• Depends on BPO model

Phase1? Is this too coupled?

10

THE WORKFLOW 11

Process & Infrastructure Compound design from

Pharma (Design)

Synthesis/ Make

Screen Compounds/

Test

Data Analysis

12

• IT system to share designs • Track metrics

• Weekly TC/ reports monitor progress

• Reagent store and database

• ELN to capture synthesis information

• Patent ready format • Bio-ELN in progress

• Test request system • Sample storage • Shipping compounds

• QC of data • IT upload system • Process to track failed

analysis

• Monitoring performance • Governance • Audits and compliance

Customer (pharma/ biotech)

Partner

Project sharing

Design sharing

Project sharing

Chemistry Synthesis Experiment

D M T A DMTA: Requesting and Tracking

Design Sharing environment

Capture of Chemical Synthesis and accessible back into Pharma

Screening data (DMPK, Biology)

Project collabo-ration spaces

Example Pre-clinical Workflow Design, Make, Test, Analyse (DMTA)

13

Screening (DMPK, Biology) Request

Screening Data to Pharma


Some Options….

14

<data exchange format>

Broker Application or translator

Shared Application (apps, Citrix, Web)

BIORULES SUPPORT ALL USERS 15

Curlew Research 2014 Flickr user sarah0s / Creative Commons Licensed

16


Business Rules – AZ Drivers

●Get visibility of our assets

●Sharing of experience

●Securing information for the longer term

●Reduced Cycle Time by not repeating work oDon't do anything already done by someone else especially if it didn't work

oDo Build on others’ learning

●Decision support oDiscovery has distributed decision making processes oEverybody makes decisions on a daily basis oNeed as much information as possible

17


Project lifetime Time

Information Value

After Project closure

Structured Information Value

Poor meta data Lack of curation process

Good information practice Clear business rules Curation process defined

Can we quantify this gap? When do investments payback?

Data created for specific Project Reasonable knowledge of data & decisions

What do decision makers need? The Customer changes over time, so rules need to adapt

http://www.b-eye-network.com/view/3365?jsessionid=48f7500a16e486668a5b968273f709e2

http://www.b-eye-network.com/blogs/linstedt/archives/2007/01/time_value_of_m.php

http://www-128.ibm.com/developerworks/webservices/library/ws-soa-ims2/

18














People

●Get the right people oYour best people are always busy oYou don't want anyone that is easily available

●Skills needed oBe able to see the “Big Picture” oKnowledgeable in their business area oGood inter-personal skills oCapable of making decisions

●Get them at the right level of organisation oMust know how the business works day-to-day

● Include all relevant people oFor us this meant representatives from 5 research areas

situated on 8 sites over 4 countries ~ 20 people!! 19


Please keep to the Path!


Training is part of knowledge management

Best practice, Minimum Information and auditing • Define Minimum information requirements

• Experiments (minimum spectra needed, use of templates for common transformations)

• Screening data (based on Assay protocol) • Reports (Standard templates)

• Auditing of data • Both internally and externally created • Peer review

• Training • Hands on training • Exams to support learning • Super users on both sides 22



Information Value increases with relationships

24

VALUE VALUE VALUE VALUE


You can ease the issues here

This is Manageable with good metadata

What type of data?

Good enterprise search can bring real value

25

Summary • The type of relationship and its length will shape

information sharing approaches • Requires a good partnership between all parties

• Not just about imposing large company ideas/tools on a small agile collaborator or CROs

• Your scientists will put a great deal of effort into collaborating, help them be part of curation • Use common business rules, agree on vocabulary • Data Curation supports good experiments

• Super-user concept, local experts • Work with your software providers for lighter weight

solutions

26

Curlew Research 2014 http://thetechnoliterate.wordpress.com/2013/04/24/its-not-just-about-the-technology/


Thanks to... Liz Calder

Eva Lotta Westberg

Janet Nason

Dave Nicholls

Vijay Chhajlani

Steve Peters

Goran Hanson

IBIS and BioELN Teams

Chris Davies David Drake Garry Pairaudeau Kyle Fang Hong Xuo Niklas Fjellman Christine Xia Barry Jones

28

And finally ….Discussion • Would standards support better data sharing? • Would common business rules help? • What technologies enable easier collaboration? • How can we structure and mange non-repetitive data

and make them searchable? (Data generated on a daily basis could with some effort be standardized and structured. These could be documented in databases and entered into tables.

• How could we capture, store and retrieve data from ad-hoc experiments that is unique in its kind?) 29
