DOCSLIB.ORG

Invited Conference Speakers

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Invited Conference Speakers Medical Research Council Conference on Biostatistics in celebration of the MRC Biostatistics Unit's Centenary Year 24th - 26th March 2014 | Queens' College Cambridge, UK Invited Conference Speakers Professor Tony Ades, School of Social and Community Medicine, University of Bristol Tony Ades is Professor of Public Health Science at the University of Bristol and leads a programme on methods for evidence synthesis in epidemiology and decision modelling. This was originally funded through the MRC Health Services Research Collaboration. With Guobing Lu, Nicky Welton, Sofia DIas, Debbi Caldwell, Malcolm Price, Aicha Goubar and many collaborators in Cambridge and elsewhere, the programme has contributed original research on Network Meta- analysis, multi-parameter synthesis models for the epidemiology of HIV and chlamydia, synthesis for Markov models, expected value of information, and other topics. Tony was a member of the Appraisals Committee at the National Institute for Clinical Excellence (NICE), 2003-2013, and was awarded a lifetime achievement award in 2010 by the Society for Research Synthesis Methodology. Seminar Title: Synthesis of treatment effects, Mappings between outcomes, and test responsiveness Tuesday 25 March, 11.00-12.30 session Abstract: Synthesis of treatment effects, mappings between outcomes, standardisation, and test responsiveness AE Ades, Guobing Lu, Daphne Kounali. School of Social and Community Medicine, University of Bristol Abstract: We shall report on some experiments with a new class of models for synthesis of treatment effect evidence on “similar” outcomes, within and between trials. They are intended particularly for synthesis of patient- and clinician- reported outcomes that are subject to measurement error. These models assume that treatment effects on different outcomes are in fixed, or approximately fixed, ratios across trials. The task is therefore to estimate the pooled treatment effect on one of the outcomes, and “mapping coefficients” between the outcomes. We illustrate several uses of this model: for estimation of pooled effects, and estimation of “mappings” that are used in health economic analysis. We also show that it is an alternative to standardisation of treatment effects by dividing them by the sample standard deviation, and that it has superior properties. Finally, the models lead naturally to a meta-analytic account of the relative responsiveness of test instruments to treatment. Professor Per Kragh Andersen, Department of Biostatistics, Institute of Public Health, University of Copenhagen Per Kragh Andersen is Professor at the Department of Biostatistics. He took his Master's degree in 1978, a PhD degree in MathematicalS tatistics in 1982, and a Dr.Med.Sci. degree in 1997 - all from University of Copenhagen. Per’s main research interests are survival/event history analysis and analysis of cohort studies and other epidemiological applications of statistics. Seminar Title: Causal inference in survival analysis based on pseudo-observations Wednesday 26 March, 09.00-10.30 session Abstract: In causal inference for survival time outcomes the target is often taken to be the survival status at some time T (or maybe at a series of time points T_1, T_2,...). Since this is incompletely observed because of censoring standard approaches use inverse probability of censoring weighting. We will use another method and calculate pseudo- observations as outcome variables for each subject at T_1, T_2 etc. and then do an "ordinary complete data causal inference". Professor Nicky Best, GlaxoSmithKline Nicky Best recently joined GlaxoSmithKline as director of the new Statistical Innovation Group. Prior to this, she was Professor of Statistics & Epidemiology, School of Public Health, at Imperial College and Senior Investigator at the MRC- PHE Centre for Environment & Health, where her research focused on development and application of Bayesian methods in the health and social sciences. She is also a co-developer of the BUGS statistical software for Bayesian analysis. She was awarded the Royal Statistical Society Guy Medal in Bronze in 2004. Seminar Title: The BUGS Project: origins, challenges, and future directions Tuesday 25 March, 13.30-15.00 session Abstract: Abstract: BUGS is a software package for Bayesian inference using Gibbs sampling. The software has been instrumental in raising awareness of Bayesian modelling among both academic and commercial communities internationally, and has enjoyed considerable success over its almost 25 year life-span. In this talk we will present a brief overview of the BUGS project, illustrating some of the key evolutionary steps via the challenging applications that inspired them. Areas considered include disease mapping, prior elicitation, non-linear regression, model uncertainty, feedback control and evidence synthesis. Join us on twitter @MRC_BSU Medical Research Council Conference on Biostatistics in celebration of the MRC Biostatistics Unit's Centenary Year 24th - 26th March 2014 | Queens' College Cambridge, UK Dr Leonardo Bottolo, Faculty of Natural Sciences, Department of Mathematics, Imperial College London Leonardo Bottolo is a Lecturer in Statistics at Imperial College. He is interested in the application of Bayesian computational methods to exploit causes of genetic variation in final and intermediate phenotypes and their interaction, i.e. complex traits and transcriptional abundance respectively. His goal is to build a class of general purpose statistical tools that, together with portable, efficient and publicly available software solutions, can be used in the future to better understand the molecular mechanisms of disease pathogenesis. Seminar Title: Multivariate mixture priors for subgroups classification Tuesday 25 March, 15.30-17.00 session Abstract: Technologies for the collection of genetic data, such as those based on RNA-Seq, are developing at a very fast rate. A crucial objective in these studies is gene clustering based on the similarity of their level of expression. A more important clustering task concerns samples themselves: this goes under the name of molecular profiling, and aims at identifying similarities across samples based on a few genes identified from the previous stage of the analysis. In this talk we present a fully Bayesian hierarchical model for molecular profiling. A key ingredient of our approach is represented by the use of mixture distributions. First of all expression measurements are assumed to originate from a three-component mixture distribution, representing the underlying population classes (baseline, underexpression and overexpression) of samples relative to gene-expression. For each gene we assume a specific (random) probability of belonging to any of the three classes. Current Bayesian modelling assumes that the collection of gene-specific probabilities is exchangeable.Exchangeability assumption exhibits some drawbacks, the main one being its inability to capture heterogeneity in gene-behavior. This led us to model the gene-specific probabilities of under- overexpression using a multivariate mixture of prior distributions with an unknown number of components: this represents the novelty of our approach which is meant to capture variation in gene behaviour and to identify clusters of genes in relation to their probability of under and overexpression. The model is applied to gene expression levels from RNA-Seq analysis of left ventricular tissue derived from a cohort comprising 33 dilated cardiomyopathy patients with end-stage heart failure who underwent left ventricular assist device implant. Of these patients, 24 subsequently recovered their function whereas 9 did not. There is no obvious distinguishing features between them at the time the sample was taken. Molecular profiling is derived to predict recovery vs non-recovery patients. This is a joint work with Petros Dellaportas and Athanassios Petralias, Dept of Statistics, Athens University. Professor Norman Breslow, Department of Biostatistics, University of Washington Emeritus Professor and former Chairman of the Department of Biostatistics at the University of Washington. Member of the Biostatistics Program, Division of Public Health Sciences, Fred Hutchinson Cancer Research Center. He was awarded the Mortimer Spiegelman Award of the American Public Health Association (1978), the Senior U.S. Scientist Award of the Alexander von Humboldt Foundation (1982-83), the 1995 George W. Snedecor Award of the Committee of Presidents of Statistical Societies (COPSS), the 1995 R.A. Fisher Award of COPSS, the 1998 Statistician of the Year awarded by the Chicago Chapter of the American Statistical Association, Docteur Honoris Causa from Bordeaux (2001), the Nathan Mantel Award, Section on Statistics in Epidemiology of the American Statistical Association (2002), was elected Honorary Fellow of the Royal Statistical Society (1994) and was elected member of the Institute of Medicine of the National Academy of Science. Professor Breslow has made pioneering contributions to the advancement of biostatistical methodological developments in survival analysis, generalized linear mixed models and the design of case-control and cohort studies. He has made instrumental contributions in cancer epidemiology and is a founding member of the National Wilms' Tumor Study Committee. He is co-author of the influential textbooks, Statistical Methods in Cancer Research (vols. I & II), with Nicholas E. Day. Seminar Title: Using the Whole Cohort in the Analysis of Case-Control Data: Applications
Load more

Recommended publications
  • Royal Statistical Scandal
    Royal Statistical Scandal False and misleading claims by the Royal Statistical Society Including on human poverty and UN global goals Documentary evidence Matt Berkley Draft 27 June 2019 1 "The Code also requires us to be competent. ... We must also know our limits and not go beyond what we know.... John Pullinger RSS President" https://www.statslife.org.uk/news/3338-rss-publishes-revised-code-of- conduct "If the Royal Statistical Society cannot provide reasonable evidence on inflation faced by poor people, changing needs, assets or debts from 2008 to 2018, I propose that it retract the honour and that the President makes a statement while he holds office." Matt Berkley 27 Dec 2018 2 "a recent World Bank study showed that nearly half of low-and middle- income countries had insufficient data to monitor poverty rates (2002- 2011)." Royal Statistical Society news item 2015 1 "Max Roser from Oxford points out that newspapers could have legitimately run the headline ' Number of people in extreme poverty fell by 137,000 since yesterday' every single day for the past 25 years... Careless statistical reporting could cost lives." President of the Royal Statistical Society Lecture to the Independent Press Standards Organisation April 2018 2 1 https://www.statslife.org.uk/news/2495-global-partnership-for- sustainable-development-data-launches-at-un-summit 2 https://www.statslife.org.uk/features/3790-risk-statistics-and-the-media 3 "Mistaken or malicious misinformation can change your world... When the government is wrong about you it will hurt you too but you may never know how.
    [Show full text]
  • September 2010
    THE ISBA BULLETIN Vol. 17 No. 3 September 2010 The official bulletin of the International Society for Bayesian Analysis AMESSAGE FROM THE membership renewal (there will be special provi- PRESIDENT sions for members who hold multi-year member- ships). Peter Muller¨ The Bayesian Nonparametrics Section (IS- ISBA President, 2010 BA/BNP) is already up and running, and plan- [email protected] ning the 2011 BNP workshop. Please see the News from the World section in this Bulletin First some sad news. In August we lost two and our homepage (select “business” and “mee- big Bayesians. Julian Besag passed away on Au- tings”). gust 6, and Arnold Zellner passed away on Au- gust 11. Arnold was one of the founding ISBA ISBA/SBSS Educational Initiative. Jointly presidents and was instrumental to get Bayesian with ASA/SBSS (American Statistical Associa- started. Obituaries on this issue of the Analysis tion, Section on Bayesian Statistical Science) we Bulletin and on our homepage acknowledge Juli- launched a new joint educational initiative. The an and Arnold’s pathbreaking contributions and initiative formalized a long standing history of their impact on the lives of many people in our collaboration of ISBA and ASA/SBSS in related research community. They will be missed dearly! matters. Continued on page 2. ISBA Elections 2010. Please check out the elec- tion statements of the candidates for the upco- In this issue ming ISBA elections. We have an amazing slate ‰ A MESSAGE FROM THE BA EDITOR of candidates. Thanks to the nominating commit- *Page 2 tee, Mike West (chair), Renato Martins Assuncao,˜ Jennifer Hill, Beatrix Jones, Jaeyong Lee, Yasuhi- ‰ 2010 ISBA ELECTION *Page 3 ro Omori and Gareth Robert! ‰ SAVAGE AWARD AND MITCHELL PRIZE 2010 *Page 8 Sections.
    [Show full text]
  • IMS Bulletin 39(4)
    Volume 39 • Issue 4 IMS1935–2010 Bulletin May 2010 Meet the 2010 candidates Contents 1 IMS Elections 2–3 Members’ News: new ISI members; Adrian Raftery; It’s time for the 2010 IMS elections, and Richard Smith; we introduce this year’s nominees who are IMS Collections vol 5 standing for IMS President-Elect and for IMS Council. You can read all the candi- 4 IMS Election candidates dates’ statements, starting on page 4. 9 Amendments to This year there are also amendments Constitution and Bylaws to the Constitution and Bylaws to vote Letter to the Editor 11 on: they are listed The candidate for IMS President-Elect is Medallion Preview: Laurens on page 9. 13 Ruth Williams de Haan Voting is open until June 26, so 14 COPSS Fisher lecture: Bruce https://secure.imstat.org/secure/vote2010/vote2010.asp Lindsay please visit to cast your vote! 15 Rick’s Ramblings: March Madness 16 Terence’s Stuff: And ANOVA thing 17 IMS meetings 27 Other meetings 30 Employment Opportunities 31 International Calendar of Statistical Events The ten Council candidates, clockwise from top left, are: 35 Information for Advertisers Krzysztof Burdzy, Francisco Cribari-Neto, Arnoldo Frigessi, Peter Kim, Steve Lalley, Neal Madras, Gennady Samorodnitsky, Ingrid Van Keilegom, Yazhen Wang and Wing H Wong Abstract submission deadline extended to April 30 IMS Bulletin 2 . IMs Bulletin Volume 39 . Issue 4 Volume 39 • Issue 4 May 2010 IMS members’ news ISSN 1544-1881 International Statistical Institute elects new members Contact information Among the 54 new elected ISI members are several IMS members. We congratulate IMS IMS Bulletin Editor: Xuming He Fellow Jon Wellner, and IMS members: Subhabrata Chakraborti, USA; Liliana Forzani, Assistant Editor: Tati Howell Argentina; Ronald D.
    [Show full text]
  • Tea: a High-Level Language and Runtime System for Automating
    Tea: A High-level Language and Runtime System for Automating Statistical Analysis Eunice Jun1, Maureen Daum1, Jared Roesch1, Sarah Chasins2, Emery Berger34, Rene Just1, Katharina Reinecke1 1University of Washington, 2University of California, 3University of Seattle, WA Berkeley, CA Massachusetts Amherst, femjun, mdaum, jroesch, [email protected] 4Microsoft Research, rjust, [email protected] Redmond, WA [email protected] ABSTRACT Since the development of modern statistical methods (e.g., Though statistical analyses are centered on research questions Student’s t-test, ANOVA, etc.), statisticians have acknowl- and hypotheses, current statistical analysis tools are not. Users edged the difficulty of identifying which statistical tests people must first translate their hypotheses into specific statistical should use to answer their specific research questions. Almost tests and then perform API calls with functions and parame- a century later, choosing appropriate statistical tests for eval- ters. To do so accurately requires that users have statistical uating a hypothesis remains a challenge. As a consequence, expertise. To lower this barrier to valid, replicable statistical errors in statistical analyses are common [26], especially given analysis, we introduce Tea, a high-level declarative language that data analysis has become a common task for people with and runtime system. In Tea, users express their study de- little to no statistical expertise. sign, any parametric assumptions, and their hypotheses. Tea A wide variety of tools (such as SPSS [55], SAS [54], and compiles these high-level specifications into a constraint satis- JMP [52]), programming languages (e.g., R [53]), and libraries faction problem that determines the set of valid statistical tests (including numpy [40], scipy [23], and statsmodels [45]), en- and then executes them to test the hypothesis.
    [Show full text]
  • Computational and Financial Econometrics (CFE 2016)
    CFE-CMStatistics 2016 PROGRAMME AND ABSTRACTS 10th International Conference on Computational and Financial Econometrics (CFE 2016) http://www.cfenetwork.org/CFE2016 and 9th International Conference of the ERCIM (European Research Consortium for Informatics and Mathematics) Working Group on Computational and Methodological Statistics (CMStatistics 2016) http://www.cmstatistics.org/CMStatistics2016 Higher Technical School of Engineering, University of Seville, Spain 9 – 11 December 2016 c CFE and CMStatistics networks. All rights reserved. I CFE-CMStatistics 2016 ISBN 978-9963-2227-1-1 c 2016 - CFE and CMStatistics networks Technical Editors: Angela Blanco-Fernandez and Gil Gonzalez-Rodriguez. All rights reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted, in any other form or by any means without the prior permission from the publisher. II c CFE and CMStatistics networks. All rights reserved. CFE-CMStatistics 2016 International Organizing Committee: Ana Colubi, Stella Hadjiantoni, M. Dolores Jimenez-Gamero, Erricos Kontoghiorghes, George Loizou and Herman Van Dijk. CFE 2016 Co-chairs: Peter Boswijk, Jianqing Fang, Alain Hecq, Michael Smith. CFE 2016 Programme Committee: Alessandra Amendola, Ansgar Belke, Monica Billio, Josep Lluis Carrion-i-Silvestre, Roberto Casarin, Veronika Czellar, Serge Darolles, Manfred Deistler, Jean-David Fermanian, Sylvia Fruehwirth-Schnatter, Marc Hallin, Matthew Harding, Jan Jacobs, Menelaos Karanasos, Daniel Kaufmann, Robert Kunst, Degui Li, Marco Lippi, Zudi Lu, Luis Filipe Martins, Gian Luigi Mazzi, Tucker McElroy, J Isaac Miller, Bent Nielsen, Yasuhiro Omori, Michael Owyang, Gareth Peters, Michael Pitt, Stephen Pollock, Tommaso Proi- etti, Artem Prokhorov, Anders Rahbek, Francesco Ravazzolo, Jeroen Rombouts, Eduardo Rossi, Christos Savva, Willi Semmler, Laura Spierdijk, Giuseppe Storti, Jonathan Stroud, Genaro Sucarrat, Jean-Pierre Ur- bain, Shaun Vahey, Helena Veiga, Carlos Velasco, Helga Wagner, Martin Wagner, Toshiaki Watanabe and Peter Winker.
    [Show full text]
  • Spatio-Temporal Point Processes: Methods and Applications Peter J
    Johns Hopkins University, Dept. of Biostatistics Working Papers 6-27-2005 Spatio-temporal Point Processes: Methods and Applications Peter J. Diggle Medical Statistics Unit, Lancaster University, UK & Department of Biostatistics, Johns Hopkins Bloomberg School of Public Health, [email protected] Suggested Citation Diggle, Peter J., "Spatio-temporal Point Processes: Methods and Applications" (June 2005). Johns Hopkins University, Dept. of Biostatistics Working Papers. Working Paper 78. http://biostats.bepress.com/jhubiostat/paper78 This working paper is hosted by The Berkeley Electronic Press (bepress) and may not be commercially reproduced without the permission of the copyright holder. Copyright © 2011 by the authors Spatio-temporal Point Processes: Methods and Applications Peter J Diggle (Department of Mathematics and Statistics, Lancaster University and Department of Biostatistics, Johns Hopkins University School of Public Health) June 27, 2005 1 Introduction This chapter is concerned with the analysis of data whose basic format is (xi; ti) : i = 1; :::; n where each xi denotes the location and ti the corresponding time of occurrence of an event of interest. We shall assume that the data form a complete record of all events which occur within a pre-specified spatial region A and a pre-specified time- interval, (0; T ). We call a data-set of this kind a spatio-temporal point pattern, and the underlying stochastic model for the data a spatio-temporal point process. 1.1 Motivating examples 1.1.1 Amacrine cells in the retina of a rabbit One general approach to analysing spatio-temporal point process data is to extend existing methods for purely spatial data by considering the time of occurrence as a distinguishing feature, or mark, attached to each event.
    [Show full text]
  • Amstat News May 2011 President's Invited Column
    May 2011 • Issue #407 AMSTATNEWS The Membership Magazine of the American Statistical Association • http://magazine.amstat.org Croatia Bosnia Serbia Herzegovina ThroughPeace Statistics ALSO: Meet Susan Boehmer, New IRS Statistics of Income Director Publications Agreement No. 41544521 Mining the Science out of Marketing Math Sciences in 2025 AMSTATNews May 2011 • Issue #407 Executive Director Ron Wasserstein: [email protected] Associate Executive Director and Director of Operations features Stephen Porzio: [email protected] 3 President’s Invited Column Director of Education Martha Aliaga: [email protected] 5 Pfizer Contributes to ASA's Educational Ambassador Program Director of Science Policy 5 TAS Article Cited in Supreme Court Case Steve Pierson: [email protected] 7 Writing Workshop for Junior Researchers to Take Place at JSM Managing Editor Megan Murphy: [email protected] 8 Meet Susan Boehmer, New IRS Statistics of Income Director Production Coordinators/Graphic Designers 10 Peace Through Statistics Melissa Muko: [email protected] 15 Statistics in Biopharmaceutical Research Kathryn Wright: [email protected] May Issue of SBR: A Festschrift for Gary Koch Publications Coordinator Val Nirala: [email protected] 16 Journal of the American Statistical Association March JASA Features ASA President’s Invited Address Advertising Manager Claudine Donovan: [email protected] 19 Technometrics Fingerprint Individuality Assessment Featured in May Issue Contributing Staff Members Pam Craven • Rosanne Desmone • Fay Gallagher • Eric Sampson Amstat News welcomes news items and letters from readers on matters of interest to the association and the profession. Address correspondence to Managing Editor, Amstat News, American Statistical Association, 732 North Washington Street, Alexandria VA 22314-1943 USA, or email amstat@ columns amstat.org.
    [Show full text]
  • December 2000
    THE ISBA BULLETIN Vol. 7 No. 4 December 2000 The o±cial bulletin of the International Society for Bayesian Analysis A WORD FROM already lays out all the elements mere statisticians might have THE PRESIDENT of the philosophical position anything to say to them that by Philip Dawid that he was to continue to could possibly be worth ISBA President develop and promote (to a listening to. I recently acted as [email protected] largely uncomprehending an expert witness for the audience) for the rest of his life. defence in a murder appeal, Radical Probabilism He is utterly uncompromising which revolved around a Modern Bayesianism is doing in his rejection of the realist variant of the “Prosecutor’s a wonderful job in an enormous conception that Probability is Fallacy” (the confusion of range of applied activities, somehow “out there in the world”, P (innocencejevidence) with supplying modelling, data and in his pragmatist emphasis P ('evidencejinnocence)). $ analysis and inference on Subjective Probability as Contents procedures to nourish parts that something that can be measured other techniques cannot reach. and regulated by suitable ➤ ISBA Elections and Logo But Bayesianism is far more instruments (betting behaviour, ☛ Page 2 than a bag of tricks for helping or proper scoring rules). other specialists out with their What de Finetti constructed ➤ Interview with Lindley tricky problems – it is a totally was, essentially, a whole new ☛ Page 3 original way of thinking about theory of logic – in the broad ➤ New prizes the world we live in. I was sense of principles for thinking ☛ Page 5 forcibly struck by this when I and learning about how the had to deliver some brief world behaves.
    [Show full text]
  • Letter from the RSS Covid-19 Task Force To
    Letter from the Royal Statistical Society Covid-19 Task Force to the Guardian regarding data transparency 7 August 2020 The Covid-19 public health crisis has placed a sharp emphasis on the role of data in government decision-making. During the current phase – where the emphasis is on Test and Trace and local lockdowns – data is playing an increasingly central role in informing policy. We recognise that the stakes are high and that decisions need to be made quickly. However, this makes it even more important that data is used in a responsible and effective manner. Transparency around the data that are being used to inform decisions is central to this. Over the past week, there have been two major data-led government announcements where the supporting data were not made available at the time. First, the announcement that home and garden visits would be made illegal in parts of northern England. The Prime Minister cited unpublished data which suggested that these visits were the main setting for transmission. Second, the purchase of two new tests for the virus that claim to deliver results within 90 minutes, without data regarding the tests’ effectiveness being published. We are concerned about the lack of transparency in these two cases – these are important decisions and the data upon which they are based should be publicly available for scrutiny, as Paul Nurse pointed out in this paper at the weekend. Government rhetoric often treats data as a managerial tool for informing decisions. But beyond this, transparency and well sign-posted data builds public trust and encourages compliance: the daily provision of statistical information was an integral part of full lockdown and was both expected and valued by the public.
    [Show full text]
  • The BUGS Project: Evolution, Critique and Future Directions
    STATISTICS IN MEDICINE Statist. Med. 2009; 28:3049–3067 Published online 24 July 2009 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/sim.3680 The BUGS project: Evolution, critique and future directions 1, , 1 2 3 David Lunn ∗ †, David Spiegelhalter ,AndrewThomas and Nicky Best 1Medical Research Council Biostatistics Unit, Institute of Public Health, University Forvie Site, Robinson Way, Cambridge CB2 0SR, U.K. 2School of Mathematics and Statistics, Mathematical Institute, North Haugh, St. Andrews, Fife KY16 9SS, U.K. 3Department of Epidemiology and Public Health, Imperial College London, St. Mary’s Campus, Norfolk Place, London W2 1PG, U.K. SUMMARY BUGS is a software package for Bayesian inference using Gibbs sampling. The software has been instru- mental in raising awareness of Bayesian modelling among both academic and commercial communities internationally, and has enjoyed considerable success over its 20-year life span. Despite this, the software has a number of shortcomings and a principal aim of this paper is to provide a balanced critical appraisal, in particular highlighting how various ideas have led to unprecedented flexibility while at the same time producing negative side effects. We also present a historical overview of the BUGS project and some future perspectives. Copyright 2009 John Wiley & Sons, Ltd. KEY WORDS: BUGS; WinBUGS; OpenBUGS; Bayesian modelling; graphical models 1. INTRODUCTION BUGS 1 is a software package for performing Bayesian inference using Gibbs sampling 2, 3 . The BUGS[ ] project began at the Medical Research Council Biostatistics Unit in Cambridge in[ 1989.] Since that time the software has become one of the most popular statistical modelling packages, with, at the time of writing, over 30000 registered users of WinBUGS (the Microsoft Windows incarnation of the software) worldwide, and an active on-line community comprising over 8000 members.
    [Show full text]
  • Design and Analysis Issues in Family-Based Association
    Emerging Challenges in Statistical Genetics Duncan Thomas University of Southern California Human Genetics in the Big Science Era • “Big Data” – large n and large p and complexity e.g., NIH Biomedical Big Data Initiative (RFA-HG-14-020) • Large n: challenge for computation and data storage, but not conceptual • Large p: many data mining approaches, few grounded in statistical principles • Sparse penalized regression & hierarchical modeling from Bayesian and frequentist perspectives • Emerging –omics challenges Genetics: from Fisher to GWAS • Population genetics & heritability – Mendel / Fisher / Haldane / Wright • Segregation analysis – Likelihoods on complex pedigrees by peeling: Elston & Stewart • Linkage analysis (PCR / microsats / SNPs) – Multipoint: Lander & Green – MCMC: Thompson • Association – TDT, FBATs, etc: Spielman, Laird – GWAS: Risch & Merikangas – Post-GWAS: pathway mining, next-gen sequencing Association: From hypothesis-driven to agnostic research Candidate pathways Candidate Hierarchical GWAS genes models (ht-SNPs) Ontologies Pathway mining MRC BSU SGX Plans Objectives: – Integrating structural and prior information for sparse regression analysis of high dimensional data – Clustering models for exposure-disease associations – Integrating network information – Penalised regression and Bayesian variable selection – Mechanistic models of cellular processes – Statistical computing for large scale genomics data Targeted areas of impact : – gene regulation and immunological response – biomarker based signatures – targeting
    [Show full text]
  • Stan User's Guide 2.27
    Stan User’s Guide Version 2.27 Stan Development Team Contents Overview 9 Part 1. Example Models 11 1. Regression Models 12 1.1 Linear regression 12 1.2 The QR reparameterization 14 1.3 Priors for coefficients and scales 16 1.4 Robust noise models 16 1.5 Logistic and probit regression 17 1.6 Multi-logit regression 19 1.7 Parameterizing centered vectors 21 1.8 Ordered logistic and probit regression 24 1.9 Hierarchical logistic regression 25 1.10 Hierarchical priors 28 1.11 Item-response theory models 29 1.12 Priors for identifiability 32 1.13 Multivariate priors for hierarchical models 33 1.14 Prediction, forecasting, and backcasting 41 1.15 Multivariate outcomes 42 1.16 Applications of pseudorandom number generation 48 2. Time-Series Models 51 2.1 Autoregressive models 51 2.2 Modeling temporal heteroscedasticity 54 2.3 Moving average models 55 2.4 Autoregressive moving average models 58 2.5 Stochastic volatility models 60 2.6 Hidden Markov models 63 3. Missing Data and Partially Known Parameters 70 1 CONTENTS 2 3.1 Missing data 70 3.2 Partially known parameters 71 3.3 Sliced missing data 72 3.4 Loading matrix for factor analysis 73 3.5 Missing multivariate data 74 4. Truncated or Censored Data 77 4.1 Truncated distributions 77 4.2 Truncated data 77 4.3 Censored data 79 5. Finite Mixtures 82 5.1 Relation to clustering 82 5.2 Latent discrete parameterization 82 5.3 Summing out the responsibility parameter 83 5.4 Vectorizing mixtures 86 5.5 Inferences supported by mixtures 87 5.6 Zero-inflated and hurdle models 90 5.7 Priors and effective data size in mixture models 94 6.
    [Show full text]