Interactive graphics in Pharma R&D: The right decision
description
Transcript of Interactive graphics in Pharma R&D: The right decision
Interactive graphics in Pharma R&D:The right decision
mg r af ti
Francois Mercier, @mgrafitStrata-EU 2013, London, 12-Nov-2013
MultipleSclerosis
1/12
2,500,000 People with MS, in the worldMainly in western countriesMainly women 35-45 y.o.
#15 in R&D spending (~2 $bn/yr)
*http://www.imshealth.com/, http://www.nationalmssociety.org/index.aspx
Price of a marketed MS treatment
30-45 k€/year/patient
Cost &Pr(Failure)
Development of a new drug in MS
300-600 M€
7-12 years
ComplexSystems
2/12
PharmaR&D
Roadmap
3/15
Target identificati
on
Lead discover
y
Lead optimizatio
n
Pre-clinical ADME
Phase I (FiH, safety
profiling)IND
Phase II
PoCEOP2
Phase III (confirmat
ory)
NDAReview MA
RegulatoryEnviro-nment
4/15
IN THE INTEREST OFALL,
STAY INSIDE THE BOX
Data analyses have to be reproducible• Source data are safely stored and data manipulation are traced• Data analyses are available in the form of programs (no GUI)
From PatientTo Data
5/15
Subject (healthy or ill)
Physician
Nurse
Data Warehous
e
Mobile devices (cardiac
function, vital signs)
Labs (biofluids, images, other
samples)
100 subjects ≈ 1-5 GB
From Data ToDecision
6/15
Data manager
Data scientist
Clinical project leader
Therapeutic area head
Clinical development head
Pharmaceutical division head
Raw data
Distillated information
ClinicalDevelop.Strategy
7/15
In clinical projects, a strategy is defined as a compromise between money
and science
€
3√𝑅𝑥
Data manager
Data scientist
Clinical project leader
Therapeutic area head
Clinical development head
Pharmaceutical division head
DecisionMakingProcess
8/15
Communicate
Collect
?
Visualize
Model
Refine
Commu-Nication
Risks
9/15
Loss of informationMute assumptions
CompromisesLack of integration
Misinterpretation
RISKS
How to mitigate these risks?
Raw data
Distillated information
We use models and graphics
Models&
Graphics
10/15
Model Visualize
Two sides of the same coin
− Mathematical language+ Assumptions
+ Universal language− Limited in space and form (2D)
*Data from Nakov et al. Am. J. Hypertension, 2002, 15:583-9.
10 30 100
Darusentan dose (mg)*
DSB
P (m
m H
g)
• Gaussian distrib⁰• Homoscedasticity• Dose is discrete
Real lifeIs
Complex
11/15
How many static graphics to depict this model?
Conclusion
14/15
• Give control to the end-user• Make scientific discussions an enjoyable experience• Provide context, perspective• Make the assumptions explicit
Interactive graphics in Pharma R&D:The right decision
To take quick and accurate decisions
AcknowLedgments
15/15
Jean Mercier (http://www.khawai.com/)
Olivier Luttringer (Novartis)
Hadley Wickham (http://had.co.nz/)
Mike Bostock (http://bost.ocks.org/mike/)