The best of both worlds Pharma R&D IT - Informatics Rudi Verbeeck Guided analytics in the hands of the SME.

Slides:



Advertisements
Similar presentations
Gakava L Roche Products Ltd., Welwyn, UK
Advertisements

Stability Studies - Evaluation of Outcomes and Development of Documentation For Regulatory Submissions Bob Seevers.
ASSESSING RESPONSIVENESS OF HEALTH MEASUREMENTS. Link validity & reliability testing to purpose of the measure Some examples: In a diagnostic instrument,
© Department of Statistics 2012 STATS 330 Lecture 32: Slide 1 Stats 330: Lecture 32.
EPIDEMIOLOGY AND GENETICS OF ALZHEIMER´S DISEASE
Clinical Decision Support: Using Logistic Regression to Diagnose COPD and CHF ©2012 Wayne G. Fischer, PhD 1 COPD patient inclusion criteria: Discharged.
Estimation and Reporting of Heterogeneity of Treatment Effects in Observational Comparative Effectiveness Research Prepared for: Agency for Healthcare.
Recursive Partitioning Method on Survival Outcomes for Personalized Medicine 2nd International Conference on Predictive, Preventive and Personalized Medicine.
Introduction to Data Mining with XLMiner
Alzheimer’s Disease Neuroimaging Initiative STEERING COMMITTEE April
CSF tau Is it an informative biomarker of AD pathology Chris Clark Alzheimer’s Disease Center University of Pennsylvania.
Lecture 6: Multiple Regression
Irwin/McGraw-Hill © The McGraw-Hill Companies, Inc., 2000 LIND MASON MARCHAL 1-1 Chapter Twelve Multiple Regression and Correlation Analysis GOALS When.
Correlational Designs
Mild Cognitive Impairment as a Target for Drug Development Steven H. Ferris, Ph.D. Silberstein Aging and Dementia Research Center New York University School.
BPT 3113 – Management of Technology
Thoughts on Biomarker Discovery and Validation Karla Ballman, Ph.D. Division of Biostatistics October 29, 2007.
T-tests and ANOVA Statistical analysis of group differences.
Quantitative Business Analysis for Decision Making Multiple Linear RegressionAnalysis.
Application of SAS®! Enterprise Miner™ in Credit Risk Analytics
Chapter 4 Research Methods
2009 Mathematics Standards of Learning Training Institutes Algebra II Virginia Department of Education.
Alzheimer’s Assessment Assessing the Cognitive-Linguistic effects of Alzheimer’s.
3 CHAPTER Cost Behavior 3-1.
Consultants: Jillian Lyon and Mary Ehlers The Impact of Atypical Antipsychotic Use on Obstructive Sleep Apnea.
Correlation.
Defining Mild Cognitive Impairment Steven T.DeKosky, M.D. Director, Alzheimer’s Disease Research Center University of Pittsburgh Pittsburgh, PA.
EDRN Approaches to Biomarker Validation DMCC Statisticians Fred Hutchinson Cancer Research Center Margaret Pepe Ziding Feng, Mark Thornquist, Yingye Zheng,
L 1 Chapter 12 Correlational Designs EDUC 640 Dr. William M. Bauer.
Preparing Written and Oral Reports Effective Communication in CH402.
Bioinformatics Brad Windle Ph# Web Site:
Biostatistics Case Studies 2007 Peter D. Christenson Biostatistician Session 3: Incomplete Data in Longitudinal Studies.
Visual Perspectives iPLANT Visual Analytics Workshop November 5-6, 2009 ;lk Visual Analytics Bernice Rogowitz Greg Abram.
Alzheimer’s Disease  Goals  To understand what dementia is  To explore causes, risk factors, symptoms, and treatments of Alzheimer’s Disease  To better.
SS440 Seminar: Unit 4 Research in Psychopathology Dr. Angie Whalen Kaplan University 1.
The Influence of Feature Type, Feature Structure and Psycholinguistic Parameters on the Naming Performance of Semantic Dementia and Alzheimer’s Patients.
MRI as a Potential Surrogate Marker in the ADCS MCI Trial
1 Inferences About The Pearson Correlation Coefficient.
Using Predictive Classifiers in the Design of Phase III Clinical Trials Richard Simon, D.Sc. Chief, Biometric Research Branch National Cancer Institute.
Model Selection and Validation. Model-Building Process 1. Data collection and preparation 2. Reduction of explanatory or predictor variables (for exploratory.
Risk assessment for VTE Dr Roopen Arya King’s College Hospital.
Data Mining BY JEMINI ISLAM. Data Mining Outline: What is data mining? Why use data mining? How does data mining work The process of data mining Tools.
Innovative Paths to Better Medicines Design Considerations in Molecular Biomarker Discovery Studies Doris Damian and Robert McBurney June 6, 2007.
Assessing Responsiveness of Health Measurements Ian McDowell, INTA, Santiago, March 20, 2001.
Ch15: Multiple Regression 3 Nov 2011 BUSI275 Dr. Sean Ho HW7 due Tues Please download: 17-Hawlins.xls 17-Hawlins.xls.
Pan-cancer analysis of prognostic genes Jordan Anaya Omnes Res, In this study I have used publicly available clinical and.
ACT476 CAPSTONE WRITING AN USER MANUAL. Developers VS Users Developers want to write code Have little time to document or write user’s manuals Users on.
Neurobiology of Dementia Majid Barekatain, M.D., Associate Professor of Psychiatry Neuropsychiatrist Isfahan University of Medical Sciences Ordibehesht.
Books Visualizing Data by Ben Fry Data Structures and Problem Solving Using C++, 2 nd edition by Mark Allen Weiss MATLAB for Engineers, 3 rd edition by.
Apolipoprotein E and Gray Matter Loss in Mild Cognitive Impairment and Alzheimer’s Disease Spampinato MV, Goldsberry G, Mintzer J, Rumboldt Z Medical University.
Alzheimer’s Disease (AD)
Slide Slide 1 Chapter 10 Correlation and Regression 10-1 Overview 10-2 Correlation 10-3 Regression 10-4 Variation and Prediction Intervals 10-5 Multiple.
Career Related Applications of Economics In Enrollment Management Matt Bogard Coordinator, Market Research Western Kentucky University.
Study Design: Making research pretty Adam P. Sima, PhD July 13, 2016
Volume 3, Issue 4, Pages (November 2017)
Quality Assurance processes
Metabolic Syndrome and Progression of Mild Cognitive Impairment
MATH-138 Elementary Statistics
Presenter: Dr. Patricio Chrem
Anastasiia Raievska (Veramed)
Multiple Regression Prof. Andy Field.
COMBINATION OF CSF PROTEIN BIOMARKERS AND BDNF, IL10 AND IL6 GENOTYPES IN EARLY DIAGNOSIS OF ALZHEIMER’S DISEASE   Mirjana Babić Leko1, Matea Nikolac Perković2,
Volume 2, Issue 1, Pages (January 2016)
Modifying Disease Course in Alzheimer's Disease
Early Dementia Distinguishing AD From MCI
Chapter 10 Correlation and Regression
New FDA Guidance on Early Alzheimer’s Disease
Presenter: Dr. Patricio Chrem
Alexandra Kuznetsova Biostatistician, Leo Pharma
A machine learning approach to prognostic and predictive covariate identification for subgroup analysis David A. James and David Ohlssen Advanced Exploratory.
Presentation transcript:

The best of both worlds Pharma R&D IT - Informatics Rudi Verbeeck Guided analytics in the hands of the SME

Outline Changing world –More data –More analysis requirements Two worlds –Technical skills –Subject matter expertise Best of both worlds –Exploratory analysis: DIAN study –Guided analytics: phase III clinical trial planning October 2014 Pharma R&D IT Informatics2

Evolution of data In the pharmaceutical industry October 2014Pharma R&D IT Informatics

October 2014Pharma R&D IT Informatics4

Data challenges for the EMIF consortium Need for efficiency enhancing workflows October 2014Pharma R&D IT Informatics5

Skills – two worlds October 2014Pharma R&D IT Informatics6 Technical skills Domain expertise

PersonalPeerPublication Data analysis & exporation DiscussionsCommunication Handshake – packaging your analysis October 2014Pharma R&D IT Informatics7 Graphs that allow a rich analysis may not convey a message very well (and vice versa)

Example: circadian rhythm gene expression October 2014Pharma R&D IT Informatics8 Expression profile of typical genes

Exploratory analysis: data interactions Overview ⇨ zoom & filter ⇨ details on demand –A table view of the raw data gives an idea of the variables and values –Create overview visualizations, investigate distributions and correlations –Use filtering to look at subsets of the data –Detailed inspection of groups, outliers, anomalies... Brushing & linking show interactions between variables October 2014Pharma R&D IT Informatics9

Example: DIAN study DIAN = Dominantly Inherited Alzheimer Network Observational study of genetically inherited early onset Alzheimer’s disease –73 families –Mutations in amyloid precursor protein, presenilin 1, 2 –Age at onset estimated from parent –> 600 variables per patient Which baseline measurements are correlated with onset of AD? October 2014Pharma R&D IT Informatics10

Data overview October 2014Pharma R&D IT Informatics11 Show raw data in a table view Overview of measurements and visits

Explore correlations between measurements October 2014Pharma R&D IT Informatics12 General overview of measurement groups Detailed correlation for selected variable(s) Individual observations

Validated statistics: guided analytics Guide the SME through a series of decisions Give freedom to explore, backtrack, what-if Ensure sound & consistent statistical methods October 2014Pharma R&D IT Informatics13

Example: clinical phase III trial planning What patient inclusion criteria should be used for the Phase III trial of an AD compound? –We expect the compound to work best in patients with mild cognitive impairment (MCI), who are on the verge of converting to Alzheimer (AD) –We expect the compound to take some time to show clinical effect. We therefore want subjects not to convert early in the trial. –Which measurements are realistic in a trial selection setting? –Data from ADNI study (Alzheimer’s Disease Neuroimaging Initiative) October 2014Pharma R&D IT Informatics14

Guided analytics In a guided analysis, advanced statistics are packaged into a wizard-like application to guide the subject matter expert through a decision process 1.How is MCI to AD conversion measured? 2.What timeframe corresponds to “early” conversion? 3.Which (combination of) baseline measurements are predictive for early vs. late conversion? Use logistic regression modelling 4.For selected baseline measures, what cut-off value should be used as a selection criterion? Use survival analysis October 2014Pharma R&D IT Informatics15

Step 1: define early conversion October 2014Pharma R&D IT Informatics16 Text fields explain decision steps Select visit that separates early from late convertors Select conversion criterion (change in diagnosis or change in clinical dementia rating) Graph shows number of early / late convertors

Step 2: find predictive covariates October 2014Pharma R&D IT Informatics17 Select baseline covariates Verify that groups are balanced Verify variables are uncorrelated Calculate logistic regression model in R Verify model diagnostics Find significant covariates by p-value Inspect ROC for full model, stepwise model and cross validation

Step 3: find cut-off values October 2014Pharma R&D IT Informatics18 Select significant covariates from previous step Verify distribution by conversion group and determine cut-off Enter cut-off value Verify time evolution of cut-off groups Kaplan-Meier plot of conversion rates by cut-off group (calculate in R)

Conclusions Usage patterns should be supported by applications or licensing model. For example, using Spotfire –Informatician / biostatistician prepares guided analysis using full client –SME follows prepared analysis to draw conclusions using WebPlayer Think about how you present the data. Your chart encoding should be easy to understand. Conclusions from a guided analysis still need to be confirmed with a statistician. Guided analytics is a good communication tool. Distribute workload. October 2014Pharma R&D IT Informatics19