b stat news - rssb.be · b-stat news editor ... the book of abstracts in electronic and in paper...
TRANSCRIPT
B - Stat News
The Newsletter of the
Belgian Statistical Society Belgische Vereniging voor Statistiek
Société Belge de Statistique
Number 58 – May 2013
Editeurs responsables/Verantwoordelijke uitgevers :
Sophie Vanbelle, Department Methodology and Statistics, Maastricht University P. Debyeplein, 1 6229 Maastricht, The Netherlands
Herbert Thijs, Center for Statistics, Hasselt University
Agoralaan - building D, 3590 Diepenbeek
THE COUNCIL OF THE SOCIETY President
Pr. Marcel Rémon, Facultés universitaires ND de la Paix, Namur
Vice-President
Pr. Stefan Van Aelst, Universiteit Gent
Secretary of the Society
Pr. Gentiane Haesbroeck, Université de Liège
Treasurer
Pr. Roel Braekers, Universiteit Hasselt
Administrators
Pr. Céline Bugli, Université Catholique de Louvain
Pr. Christophe Croux (Katholiek Universiteit Leuven)
Dr. Filip De Ridder, Janssen Pharmaceutica
Pr. Tetyana Kadankova, Vrije Universiteit Brussel
Pr. Luc Lebrun, SPF Economie
Pr. Davy Paindaveine, Université Libre de Bruxelles
Pr. Christian Ritter, Ritter and Danielson Consulting and
Université Catholique de Louvain
Pr. Herbert Thijs, Universiteit Hasselt
Pr. Anneleen Verhasselt, Universiteit Antwerpen
Pr. Rainer Von Sachs, Université Catholique de Louvain
Website of the Society
www.sbs-bvs.be
B-Stat News editor
Sophie Vanbelle : [email protected]
Herbert Thijs: [email protected]
Webmaster
Laurence Seidel : [email protected]
2
TABLE OF CONTENTS
Adolphe Quetelet in An American Statistical Association Hall of Fame ........ 4
21th Meeting of the Belgian Statistical Society ............................................... 5
2nd VOC conference .......................................................................................... 6
European Statistical Meeting: Health Technology Assessment ................. 7
4th Nordic-Baltic Biometric conference .......................................................... 8
An Introduction to the Joint Modeling of Longitudinal and
Survival Outcomes, with applications in R .............................................. 9
15th Applied Stochastic Models and Data Analysis International
conference .................................................................................................. 10
4th Channel Network Conference ................................................................. 11
Model Selection and Nonparametric and Dependence Modeling........... 12
Conference of the International Federation of Classification Societies ... 13
Emerging Topics for Statistical Methodology in Clinical
Drug Development .................................................................................... 14
Structural Inference in Statistics ................................................................... 15
European Statistical Meeting: Structured Benefit-Risk Assessment ........ 17
3
Functional and Complex Structure Data Analysis ..................................... 18
Dynamic Predictions for Repeated Markers and Repeated Events ......... 20
2013 Nonclinical Biostatistics Conference ................................................... 21
International Hexa-Symposium on Biostatistics, Bioinformatics
and Epidemiology ..................................................................................... 22
Applications of Modeling and Simulation in Drug Development .......... 23
Forthcoming statistical events ....................................................................... 24
Recent PhD thesis ........................................................................................... 25
Job market ........................................................................................................ 27
4
ADOLPHE QUETELET IN
AN AMERICAN STATISTICAL ASSOCIATION (ASA)
HALL OF FAME Stephen M. Stigler writes in the March 2013 issue of Amstat News (pages 34-35), under the title An ASA Hall of Fame:
"On the occasion of this International Year of Statistics, and in anticipation of the 175th anniversary of the ASA in 2014, I offer a list of 20 past ASA members who where influential in bringing us to this point in our history."
Adolphe Quetelet appears prominently in this list, along with George Snedecor, Jerzy Neyman, John Tukey, W. Edward Deming, William G. Cochran, Gertrude Cox, Harold Hotelling and many others. Stigler further specifies:
"Adolphe Quetelet. Belgian, founder of the International Statistical Congresses. Played a direct role in starting the Royal Statistical Society, and by agreeing to be its first foreign member, he helped the ASA gain international recognition." Quetelet is often better known abroad than in his own country! Other examples of international recognition of Quetelet can be found in Adolphe Quetelet revisité (www.dagnelie.be/docpub/dagnelie-2005b.pdf, pages 25-28).
5
21TH MEETING OF THE BELGIAN STATISTICAL SOCIETY
Ghent (Belgium), 9-11 October 2013
Venue of the Conference: The conference will be held at the Culture and Congress center of Ghent Univeristy: Het Pand. This unique building is a former Dominican Monastry, situated beside the river Leie in the historic center of the city of Ghent.
Invited Speakers
Dr. Bradley Jones, SAS Institute, New Methods in the Design of Computer Experiments with Alternatives for Analysis
Prof. Thomas Nichols, University of Warwick, Spatial Bayesian Point Process Modeling for Neuroimaging Data
Prof. Rainer von Sachs, Université Catholique de Louvain, Shrinkage Estimation of the Dependence Structure of High-Dimensional Time Series
Prof. Tim Friede, University Medical Center Göttingen, Internal Pilot Study Designs for Clinical Trials: An Update on Recent Developments
Prof. Arnošt Komárek, Charles University Prague, Clustering for Multivariate Continuous and Discrete Longitudinal Data
Important dates 17 May 2013 Start of reduced fee registration
28 June 2013 Closing of abstract submission
15 July 2013 Notification of abstract acceptance (for oral or poster presentation)
16 August 2013 Closing of reduced fee registration
More information: http://www.bss2013.ugent.be/
6
2ND VOC CONFERENCE
Antwerpen (Belgium), 31 May 2013 The VOC Conference 2013 will be held on 31 May in Antwerp (EcoHuis). The keynote speaker for this year will be Prof. dr. Stijn Vansteelandt (Ghent University and the London School of Hygiene and Tropical Medicine). The title of his talk is “On Confounding, Prediction and Efficiency in the Analysis of Longitudinal and Cross-sectional Clustered Data”. As last year, there also will be sessions in which members and non-members present their work. The venue of the conference will be the EcoHuis in Antwerp (Turnhoutsebaan 139, 2140 Borgerhout). More information can be found here. Registering yourself can be done at: https://ppw.kuleuven.be/english/mesrg/voc-conference-2013/registration-form-for-the-second-voc-conference.
Please register before May 17th.
7
EUROPEAN STATISTICAL MEETING (EFSPI/PSI)
HEALTH TECHNOLOGY ASSESSMENT
Basel (Switzerland), 4 June 2013
Health Technology Assessment (HTA) studies the medical, social, ethical and economic implications of a health technology to a health care system, and today it is increasingly critical for manufacturers to provide additional evidence apart from that which is mandated by regulators. This seminar will invite representatives with strong statistical backgrounds from academia, HTA bodies/regulators and industry working closely in this area. They will highlight, together with case studies, some of the key challenges in performing proper HTA research as well as where statisticians and others working in this field should be involved in helping to provide analytical and methodological leadership when establishing the evidence base that will meet the needs of both regulators and payers.
More information: http://www.efspi.org/index.php?p=EFSPI+activities&fid=19
8
4TH NORDIC-BALTIC BIOMETRIC CONFERENCE
Stockholm (Sweden), 10-12 June 2013
The Nordic-Baltic Region of the International Biometric Society invites you to the 4th Nordic-Baltic Biometric Conference, NBBC13, June 10-12, to be held in Stockholm, Sweden. We welcome the submission of abstracts for oral and poster presentations on any topic of interest to members of the International Biometric Society. We are pleased to announce that the following researchers have accepted invitations to speak:
Magne Aldrin, Department of Statistical Analysis, Pattern Recognition and Image Analysis (SAMBA), Norwegian Computing Center, Norway.
Vanessa Didelez, School of Mathematics, University of Bristol, UK.
Mike Kenward, Department of Medical Statistics, London School of Hygiene & TropicalMedicine, UK.
Andrew Morris, Nuffield Department of Clinical Medicine, University of Oxford, UK.
Hannu Oja, Tampere School of Health Sciences, University of Tampere, Finland.
Andrea Rotnitzky, Harvard School of Public Health, Harvard University, USA.
Thomas Scheike, Department of Public Health, University of Copenhagen, Denmark.
Nuala Sheehan, Departments of Health Sciences and Genetics, University of Leicester, UK.
Sara Sjöstedt de Luna, Department of Mathematics and Mathematical Statistics, Umeå University, Sweden.
Aila Särkkä, Department of Mathematical Statistics, Chalmers University of Technology, Sweden
More information: http://nbbc13.org/.
9
AN INTRODUCTION TO THE JOINT MODELING OF
LONGITUDINAL AND SURVIVAL OUTCOMES, WITH
APPLICATIONS IN R
Rotterdam (The Netherlands), 18-19 June 2013 In follow-up studies different types of outcomes are typically collected for each subject. These may include several longitudinally measured responses (e.g., biomarkers, blood values), and the time until an event of particular interest occurs (e.g., death, dropout from the study). Often these outcomes are separately analyzed, but in many occasions it is of scientific interest to study their association. This type of research questions has given rise in the, very popular nowadays, class of joint models for longitudinal and time-to-event data. These models constitute an attractive paradigm for the analysis of follow-up data that is mainly applicable in two settings: First, when focus is on a survival outcome and we wish to account for the effect of endogenous time-dependents covariates measured with error (e.g., biomarkers), and second, when focus is on the longitudinal outcome and we wish to correct for non-random dropout. This two-day course is aimed at applied researchers and graduate students, and will provide a comprehensive introduction into this modeling framework. In particular, we will explain when these models should be used in practice, which are the key assumptions behind them, and how they can be utilized to extract relevant information from the data. Emphasis will be given on applications, and after the end of the course participants will be able to define appropriate joint models to answer their questions of interest. In terms of software, we will use R and illustrate how these models can be fitted using package JM. Participants will be expected to bring their own laptop computers to the session, and to have recent versions of R (http://www.r-project.org/) and of R package JM (http://cran.r-project.org/package=JM) already installed on these computers. All necessary computer code will be provided during the two afternoon computer lab sessions, and the presence of teaching assistants will ensure adequate assistance for those who become lost during the presentation.
Registration is only effective upon receipt of payment. To register please fill in the registration form and send it to: [email protected]
10
15TH APPLIED STOCHASTIC MODELS AND DATA
ANALYSIS INTERNATIONAL CONFERENCE
Mataró (Spain), 25-28 June 2013
The Applied Stochastic Models and Data Analysis International Conference (ASMDA) main aim is to welcome papers, both theoretical or practical, presenting new techniques and methodologies in the broad area of stochastic modeling and data analysis. An objective is to use the methods proposed for solving real life problems by analyzing the relevant data. Also, the use of recent advances in different fields will be promoted such as for example, new optimization and statistical methods, data warehouse, data mining and knowledge systems, computing-aided decision supports and neural computing.
Particular attention will be given to interesting applications in engineering, productions and services (maintenance, reliability, planning and control, quality control, finance, insurance, management and administration, inventory and logistics, marketing, environment, human resources, biotechnology, medicine, ...).
The publications of the conference include: The Book of Abstracts in Electronic and in Paper form Electronic Proceedings in CD and in the web in a permanent website Publications in International Journals
For more information and Abstract/Paper submission and Special Session Proposals please visit the conference website at: http://www.asmda.es and submit your contributions to [email protected]
Kind regards, Prof. Raimondo Manca University of Rome "La Sapienza" Italy Prof. Christos H. Skiadas Technical University of Crete, Greece Prof. Vladimir Zaiats Escola Politècnica Superior, Universitat de Vic, Spain
11
4TH CHANNEL NETWORK CONFERENCE
St Andrews (Scottland), 3-5 July 2013
This conference will be hosted at the University of St Andrews. The conference will start at lunchtime on Wednesday 3rd July and finish at lunchtime on Friday 5th July, with pre-conference workshops taking place on the Wednesday morning. Limited university accommodation is available from Tuesday to Saturday. Contributed talks and posters are sought.
Keynote speaker Geert Verbeke (KU Leuven)
Fisher Memorial Lecture David Spiegelhalter (Cambridge) Pre-conference workshops Parameter Redundancy, Diana Cole (University of Kent) Current Methods in Mixed Modelling, Peter Diggle (University of Lancaster)
Invited sessions Statistical Ecology
David Borchers (University of St Andrews) Eleni Matechou (Oxford University) Jérôme Dupuis (Université Paul Sabatier, Toulouse)
Mixed Modelling
Simon Wood (Bath University) Janine Illian (University of St Andrews) Adeline Samson (Université Paris Descartes)
Advances in Genomics
Mark van de Wiel (VU University) Hae Won Uh (Leiden University) Hongzhe Li (Pennsylvania University)
For further information visit http://bir.biometricsociety.org/events/channelnetworkconference or contact [email protected]
12
MODEL SELECTION AND NONPARAMETRIC AND
DEPENDENCE MODELING
Rennes (France), 8-9 July 2013
This is the first workshop of the Scientific Research Network on Asymptotic Theory for Multidimensional Statistics (FWO-Flanders, Belgium). Invited talks given by:
Claudia Czado (Technische Universität München, Germany): Pair copula constructions and applications;
Aurore Delaigle (University of Melbourne, Australia): Nonparametric regression with homogeneous group testing data;
Christian Genest (McGill University, Canada): A fresh look at extreme-value dependence modeling;
Bruce Hansen (University of Wisconsin, USA): Model averaging, sieve regression, and minimax efficiency;
Johanna Neslehova (McGill University, Canada): Copula-based tests of independence for discrete or mixed data.
Scientific organizers: G. Claeskens en I. Gijbels (KU Leuven); A. Verhasselt and P. Janssen (UHasselt) Local organizer: V. Patilea (CREST-ENSAI, Rennes) More information: http://atms.ensai.fr/
13
CONFERENCE OF THE INTERNATIONAL FEDERATION
OF CLASSIFICATION SOCIETIES 2013
Tilburg (The Netherlands), 14-17 July 2013 The 2013 conference of the International Federation of Classification Societies (IFCS) will be held 14-17 July at Tilburg University, The Netherlands. The conference theme will be ‘United through Ordination and Classification’. On July 14, preconference workshops will be held. The conference itself will start on July 15 in the morning, and will close on July 17 with a full day conference program and a conference dinner. The conference will include a president’s invited session and a presidential address, key note presentations, invited presentations, invited and contributed symposia, and oral and poster presentations. The opening session will further feature the presentation of the Chikio Hayashi Awards, which are scientific prizes for young researchers with promising track records in the areas of classification and data analysis, as a support of their professional career; for 2013 the members of the Awards Committee are: Michel Wedel (chair), Edwin Diday, Sylvia Frühwirth-Schnatter and James Ramsay. More information: http://www.tilburguniversity.edu/about-tilburg-university/schools/socialsciences/organization/departments/methodology-statistics/ifcs2013/
14
EMERGING TOPICS FOR STATISTICAL METHODOLOGY
IN CLINICAL DRUG DEVELOPMENT
Paris (France), 16-17 September 2013 This meeting will focus on the latest progress in selected statistical topics of high relevance to clinical drug development. An introduction will be given by Carl PECK, MD, Professor of Bioengineering and Therapeutic Sciences, University of California (San Francisco), and Chairman, NDA Partners LLC. The Scientific Committee, chaired by Pr Frank BRETZ from Novartis, has identified four areas of research that may change the face of clinical drug development in the near future:
Dose Finding
Subgroup Analysis &Identification
Quantifying Benefit-Risk
Graphical Approaches Contributed papers will be welcome for an oral presentation if they are relevant to any of these four areas, or as a poster presentation if they describe unrelated but innovative research on clinical drug development. All invited and selected papers may be submitted for publication in a special issue of Statistics In Medicine. All information will be available on the Web site: http://www.biopharma2013-sfds.fr
15
STRUCTURAL INFERENCE IN STATISTICS
Potsdam (Germany), 17-19 September 2013
The aim of the conference is to bring together and foster exchanges between experts of mathematical statistics and of different neighboring fields relevant to structural inference. The combination of theoretical, methodological and applied aspects also aims at identifying possible new seminal directions of research. The conference will feature two keynote mini-lecture series by Sanjoy Dasgupta (University of California, San Diego) on combining algorithmic theory with geometry and statistics, and Stéphane Mallat (École Polytechnique, Paris) on approaches linked to signal processing and approximation theory. Confirmed invited speakers are
Juditsky (Université Joseph Fourier, Grenoble) (tentatively confirmed),
G. Kerkyacharian (Université Pierre et Marie Curie, Paris),
G. Lecué (CNRS and Université Paris-Est Marne-la-Vallée),
M. Low (Wharton, University of Pennsylvania),
Munk (Georg-August-Universität Göttingen),
Nobel (University of North Carolina at Chapel Hill),
S. Tsybakov (CREST-ENSAE, Paris) (tentatively confirmed) and
M. Wegkamp (Cornell University, Ithaca) (tentatively confirmed). Further contributed talks and posters will present recent research
developments. Participants are invited to submit abstracts by May 24, 2013 (extended deadline) for presentation of their research in the form of a poster. A limited number of abstracts will be selected by the program committee for contributed talk presentation. The scientific committee is composed of the members of the DFG supported research group "Structural Inference in Statistics". Structural inference in statistics consists in estimating or adaptively exploiting some partially unknown mathematical structure underlying the observed data, in order to make the inference more efficient. Certainly, elementary structural assumptions have always been at the heart of traditional statistical modeling parametric as well as nonparametric (for instance, a simple structural model is the unknown regularity of a target function).
16
In the recent years, increasingly elaborate and diverse forms of structure have been considered, and in this movement ideas from various other fields of mathematics incorporated into statistical thinking - for instance geometry, graph theory, signal processing, approximation theory, random matrix theory, or optimization. Local organizers:
Gilles Blanchard, Natalie Neumeyer, Sonja Neiße, Franziska Göbel; Rui Wang-Müller
17
EUROPEAN STATISTICAL MEETING (EFSPI/PSI)
STRUCTURED BENEFIT-RISK ASSESSMENT
London (United Kingdom), 17 September 2013 Structured Benefit-Risk assessments on both novel and marketed medicines are now being carried out with greater frequency by the pharmaceutical industry, academia, regulatory agencies and payers in order to help with greater transparency but also to help enable better decisions to be made. This one day, face to face, meeting is being organised jointly between EFSPI and PSI to give the latest information around structured Benefit-Risk assessments from academic, industry and regulatory perspectives. The day will start with the thinking behind structured benefit risk assessments and then move onto the reality of carrying one out using examples from both industry and the IMI PROTECT Benefit Risk-Work Package. So whether you are new to Benefit-Risk, or trying to find better ways of carrying out a Benefit-Risk assessment, this will be for you.
More information: http://www.efspi.org/index.php?p=EFSPI+activities&fid=19
18
FUNCTIONAL AND COMPLEX STRUCTURE DATA
ANALYSIS
Castro Urdiales Santander (Spain), 17-21 September 2013
The ECAS 2013 session will give a general overview of the methodological and practical aspects of Functional Data Analysis (FDA) with recent advances. Functional data analysis is a branch of statistics that analyzes data providing information about curves, surfaces or anything else varying over a continuum. The continuum is often time, but may also be spatial location, wavelength, probability, etc.
The data may be so accurate that error can be ignored, may be subject to substantial measurement error, or even have a complex indirect relationship to the curve that they define.
Models for functional data and methods for their analysis may resemble those for conventional multivariate data, including linear and nonlinear regression models, principal components analysis, and many others. But the possibility of using derivative information greatly extends the power of these methods, and also leads to purely functional models such as those defined by differential equations.
Most recent developments will be considered too, and their practical approach will receive attention using the R software. For example, measurements of the heights of children over a wide range of ages have an error level so small as to be ignorable for many purposes, but daily records of precipitation at a weather station are so variable as to require careful and sophisticated analyses in order to extract something like a mean precipitation curve. These curves are often smooth in functional data analysis. In particular, functional data analyses often make use of the information in the slopes and curvatures of curves, as reflected in their derivatives. Plots of first and second derivatives as functions of t, or plots of second derivative values as functions of first derivative values, may reveal important aspects of the processes generating the data. As a consequence, curve estimation methods designed to yield good derivative estimates can play a critical role in functional data analysis.
The analysis of complex data types that is an extension of Functional Data Analysis where one considers methods to analyze data samples of complex
19
objects. Modern science is generating a need to understand, and statistically analyze, populations of increasingly complex types.
It is assumed that the participants benefit of a good background in statistics up to and including multivariate analysis, and have been exposed to matrix algebra. Participants will learn modeling real data – eventually on their own WIFI compatible laptop - but no previous acquaintance with one of the currently available software packages in FDA is required.
Important deadlines Early registration fees: 15 June 2013 Poster and short communications submission: 15 June 2013
More information: http://eio.usc.es/pub/ecas2013/index.php/registration
20
DYNAMIC PREDICTIONS FOR REPEATED MARKERS
AND REPEATED EVENTS
Bordeaux (France), 10-11 October 2013
Over two days, this workshop will provide an opportunity for epidemiologists, biostatisticians and mathematicians, actors and researchers in the field of health to learn about innovative methods of data modeling, and the cancer prediction tools available to prevent the events of patients with cancer.
Prediction models are used more and more to complement clinical reasoning, especially in the field of cancer. For this reason, the statistical models developed must provide accurate estimates and validated predicted probabilities on individuals targeted. However, even if the statistical and mathematical models have been widely developed and are still growing, methods of validating these approaches still require extensive research.
Experts in the field have been solicited and have responded favorably. An important place is also left to the posters. The objective of this workshop, without registration fee for participants will be also to intensify the exchange between all actors in the data analysis of cancer: biostatisticians, mathematicians, but also epidemiologists and clinicians. This is the hope of the organizers of this workshop, who are associated for the first time on this topic.
21
2013 NONCLINICAL BIOSTATISTICS CONFERENCE
Villanova (USA), 15-17 October 2013 The third U.S. conference dedicated entirely to nonclinical biostatistics topics will take place October 15 - 17, 2013, at the Connelly Center on the campus of Villanova University. Members of the nonclinical/preclinical Statistics community are invited to submit proposals for presentations and posters discussing significant scientific and regulatory issues. Attendees will have ample opportunity to network, share experiences and discuss current scientific issues with colleagues and leaders in the field. The Biopharm section of the American Statistical Association is a contributing sponsor of the conference. Abstract submission and registration is through the conference website
www.ncb2013.org.
Contributed presentation abstract/poster submission cutoff date is June 10, 2013.
Program
Keynote Speaker Douglas Throckmorton MD, Deputy Director, CDER, FDA
Featured Speaker Marie Davidian PhD (ASA President), North Carolina State University
Choice of half-day short course:
Strategies for Accelerating Formulation Development (Ron Snee, Snee Associates)
Applied Bayesian Statistics for Non-Clinical Areas: From Theory to Examples with Programming (Bruno Boulanger, Arlenda).
22
INTERNATIONAL HEXA-SYMPOSIUM ON
BIOSTATISTICS, BIOINFORMATICS AND
EPIDEMIOLOGY
Hasselt (Belgium), 14-15 November 2013
In celebration of
the 40th anniversary of Hasselt University, the 25th anniversary of the Master in Statistics program, the 20th anniversary of the Interuniversity Course Program, the 15th anniversary of the Center for Statistics the 5th anniversary of I-BioStat, and, last but not least, the International Year of Statistics 2013,
The Interuniversity Institute for Biostatistics and statistical Bionformatics (I-
BioStat) is proud to present the International Hexa-Symposium on
Biostatistics, Bioinformatics, and Epidemiology that will take place on November 14 and 15, 2013, at Hasselt University, Campus Diepenbeek, Building D, 3590 Diepenbeek The two-day workshop will consist of invited lectures and contributed talks.
The list of confirmed invited speakers includes: Keith Baggerly (MD
Anderson Cancer Center), Marc Buyse (IDDI), Sir David Cox (Nuffield
College), Luc Duchateau (Ghent University), Emmanuel Lesaffre
(Erasmus MC), Ingrid Van Keilegom (Catholic University of Louvain-la-
Neuve), Geert Verbeke (Catholic University of Leuven), and Joost Weyler (University of Antwerp). The workshop will conclude with an academic session, during which a honorary doctorate will be awarded. Please, book the dates in your calendar. More details about the program and organization will be provided in June.
www.uhasselt.be/hexa-symposium
23
APPLICATIONS OF MODELING AND SIMULATION IN
DRUG DEVELOPMENT
Hasselt (Belgium), 12-13 December 2013 At the end of the International Year of Statistics, the Center of Statistics at the University Hasselt organizes a workshop “Applications of Modeling and Simulation in Drug Development” as a closing event with the help of individuals from the Quantitative Sciences Department within Janssen Pharmaceutica NV. This workshop aims to bridge the gap between statistics and two disciplines that play a key role in current drug development: Systems Biology and Quantitative Pharmacology. Recenty, the term “pharmacometrics” has been coined to describe the intersection of these three areas. Historically, statistical modeling and pharmacological modeling (PK, PK/PD) approaches evolved in isolation. This workshop will feature an introduction to pharmacological modeling concepts, appropriate statistical methodology as well as state of the art applications that span the entire drug development process. The target audience consists of biostatisticians and pharmacometric modelers from both academia and industry Invited speakers confirmed their participation are:
Don Mager, U Buffalo (USA)
Geert Molenberghs, U Hasselt (Belgium)
Adrian Dunne, Janssen Pharmaceutica (Ireland)
France Mentré, Inserm Ile de France (France)
Mat Karlsson, Uppsala University (Sweden)
Philippe Jacqmin, Exprimo (Belgium)
Bart Ploeger, LAP&P (Netherlands)
Stefaan Rossenu, Merck (Belgium) You will receive more details about the workshop later on this year, Ziv Shkedy, An Vermeulen, Filip De Ridder, Adrian Dunne, Tom Jacobs
24
FORTHCOMING STATISTICAL EVENTS
May 21-23, 2013 - Rotterdam (The Netherlands), BAYES 2013 - Registration
form
June 4-6, 2013 - UCLouvain, Econometrics of Mixed Data Sampling (MIDAS)
regressions and related methods (Prof. E. Ghysels, UNC at Chapel Hill)
June 10-12, 2013 - Chicago (USA), – 2nd
International Conference and
Exhibition on Biometrics & Biostatistics (Biometrics - 2013) hosted by OMICS
Group Conferences
June 25-28, 2013 - Mataro (Spain), 15th applied stochastic model and data
analysis international conference - Registration form
July 3-2, 2013 - University St Andrews (UK), 4th Channel Network Conference
July 8-9, 2013 - Rennes (France), Workshop "Model Selection and
nonparametric and dependance modeling"
July 14-17, 2013 - Tilburg (The Netherlands), IFCS 2013 (Conference of the
International Federation of Classification Societies) - Registration form
August 25-30, 2013 - Hong Kong (China), 59th World Statistics Congress
September 17-19, 2013 - Potsdam (Germany), Conference on "Structural
inference"
September 17-21, 2013 - Castro Urdiales (Spain), Functional and Complex
Structure Data Analysis - 14th Course in the ECAS Programme
September 30 - October 1-2, 2013 - UCLouvain, High dimensional
Econometrics (Prof. J. Fan, Princeton)
October (9) 10-11, 2013 - Ghent (Belgium), 21th meeting of the Belgian
Statistical Society
October 10-11, 2013 - Bordeaux, Dynamic predictions for repeated markers
and repeated events:models and validation in cancer
November 14-15, 2013 - UHasselt, International Hexa-Symposium on
Biostatistics, Bioinformatics, and Epidemiology
December 12-13, 2013 - UHasselt, Workshop "Applications of Modeling and
Simulation in Drug Development"
25
RECENT PHD THESIS
Université de Liège (ULg)
Anne-Françoise Donneau. Contribution to the statistical analysis of incomplete longitudinal ordinal data – Promotors. Pr. A. Albert (ULg) and Pr. G. Haesbroeck (ULg) Instead of viewing ordinal variables as fully-fledged variables, some researchers consider them as being either nominal or quantitative. Therefore, when analysing data, they generally apply methods that are inappropriate because they ignore the ordinal feature of the variables. Methods for analysing ordinal outcome variables, in particular in longitudinal settings with missing data, are the main focus of this thesis. In the first part of this work, we consider the analysis of ordinal outcome data in situations where assessments are made only once for each subject. After providing a definition of an ordinal variable, the different ways to assess its relationship with a set of covariates are investigated. In this perspective, we focus on the well-known proportional odds regression model and on its strict underlying assumption. Some authors developed numerical and graphical methods to test the proportional odds assumption. When this assumption fails for some or for all the covariates, fitting a more general model, such as the partial proportional odds model or the non-proportional odds model, can be the solution. Next, our investigations on the analysis of ordinal outcome variables move to the longitudinal setting and the associated unavoidable problem of missing data. In this context, the marginal model framework was considered with the generalized estimating equations (GEE), popular for the analysis of non-Gaussian correlated data. After an adaptation of the GEE to the ordinal outcome framework, we proposed to handle the presence of missing data by considering multiple imputation (MI) prior to the data analysis. Even if not strictly speaking appropriate for ordinal data, it is a common practice for researchers to impute ordinal missing data using Markov Chain Monte Carlo (MCMC) methods designed for continuous ones, for example the multivariate normal imputation (MNI) method.
26
There is a need however to use more appropriate MI approaches specifically designed for ordinal data. In this context, we introduced the ordinal multiple imputation method (OIM) based on the proportional odds model. Then, we conducted a comprehensive simulation experiment in which we investigated the effect of several factors on the estimation of the parameters of the proportional odds model. These factors included the number of categories of the ordinal outcome, the sample size, the number of time points, the rate of missingness, the type of missingness (monotone or non-monotone) and the form of the ordinal data distribution (well-balanced or skewed). Our work shows that, whatever the pattern of missingness, the estimates derived under the MNI are highly biased, while those obtained under the OIM are almost unbiased even for datasets with a high proportion of missing data. In fact, we found that the MNI approach markedly modifies the underlying distribution of the ordinal data as opposed to the OIM. In the final part of our work, we addressed the problem of testing the proportional odds assumption for incomplete longitudinal ordinal data. Under the MNI, the type I error rate was significantly higher than the 5% nominal level while the power was markedly increased as compared to that of the full dataset (absence of missing data). Under OIM, the opposite picture occurred with lowered type I error and loss of power. As a general conclusion, our findings demonstrate that the widely used MI method based on multivariate normality in the analysis of incomplete longitudinal ordinal data does severely bias estimates of the proportional odds regression coefficients. Furthermore, multiple imputation methods, MNI or OIM, are inadequate to test the proportional assumption since they modify the distribution of the data in favor or against this popular assumption. The test based on the so-called “complete-case” dataset, i.e. eliminating all subjects with missing observations, did actually perform well despite a loss of power, at least when the rate of missingness was moderate. Throughout this work, methods were applied to real life datasets and more particularly to quality of life data of an EORTC cancer clinical trial which motivated the present research work.
27
JOB MARKET
There is an extensive list with new job offers which would be time and space consuming to publish here. Please note thet the complete list with further details is shown at our website: http://www.sbs-bvs.be/
28
EDITORIAL NOTE
We would like to publish in this Newsletter any statistical matter such as :
– information about universities, institutes (1 to 3 pages);
– lists of recent publications and technical reports;
– abstracts of recent PhD theses;
– news of members;
– forthcoming statistical events and announcements;
– short papers about teaching methods in statistics, statistics in
the industry, official statistics, etc.
Suggestions are welcome: please, contact us.
Suitable information for the next issue, prepared as (LA)TEX or WORD
FILES, should reach the editors of the Newsletter BEFORE August 30, 2013,
preferable by e-mail to:
[email protected] or [email protected]
Any change of job, address, phone number, ... ?
Please notify the Secretary of the Society:
Gentiane Haesbroeck:
Université de Liège
Institut de mathématique
Sart Tilman B37
B-4000 Liège