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 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 toa 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 , 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 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 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,

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 to the Women’s Health Initiative Trials Monday 25 March, 11.30-13.00 session

Abstract: Standard analyses of data from case-control studies that are nested in a large cohort ignore information available for cohort members not sampled for the sub-study. This paper reviews several methods designed to increase estimation efficiency by using more of the data, treating the case-control sample as a two or three phase stratified sample. The methods include in particular recent proposals by Scott and Wild (Can J Statist, 2011) for enhanced estimating equations based on models for the probability of inclusion in the case-control sample. When applied to a study of coronary heart disease among women in the hormone trials of the Women’s Health Initiative, modest but increasing gains in precision were observed depending on the amount of cohort information used in the analysis.

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

Professor Kar Keung Cheng, Public Health, Epidemiology and Biostatistics, School of Health and Population Sciences, University of Birmingham

Kar Keung Cheng, known as KK, is Professor of Public Health and Primary Care at the University of Birmingham. His main interests are in the epidemiology, prevention and control of important non-communicable diseases. He has also been closely involved in the development of primary care in China, and is Founding Head of General Practice at Peking University.

Seminar Title: Learning from the best Monday 25 March, 16.00-17.30 session

Abstract: Unlike other presentations in this Conference, this presentation will not be focusing on statistical matters. Instead, KK is going to describe how his time as a PhD student of Nick Day at the MRC Biostatistics Unit in 1990-92 has guided him in his role as an epidemiologist and health service researcher. He would reflect on how the experience of learning from the best in methodological areas has helped him to cope with the rapidly changing world of health research in the last two decades.

Professor Nicholas Edward Day

Professor Nick Day, CBE, FRS, worked at the International Agency for Research on Cancer in Lyon from 1969 to 1986, where he rose to become head of the Unit of Biostatistics and Field Studies. He was director of the Medical Research Council Biostatistics Unit from 1986 to 1989, and continued as honorary director until 1999. From 1997 until his retirement in 2004 he was co-director of the Strangeways Research Laboratory in Cambridge. During his time there, he was instrumental in influencing its development as a centre for genetic epidemiology. His personal research focus was centred upon screening and prevention. He was also professor of public health at the from 1989 to 1999, and professor of epidemiology from 1999 until 2004. Day was made a Commander of the Order of the British Empire in the 2001 New Year Honours for services to statistics and epidemiology underpinning cancer biology.

Seminar Title: The MRC Biostatistics Unit - from the past into the future Monday 25 March, 11.30-13.00 session

Abstract: A forerunner of the MRC was established under the National Insurance Act 1911, part of which was aimed at providing a degree of health care across the population. Tuberculosis was to be the major focus of health care provision, including funding for sanatoria. One old penny per year for each person insured was to be paid by central government for this purpose 'provided that the Insurance Commissioners may retain the whole or any part of the sum so payable for the purposes of research'. One short phrase in a document of several hundred pages, and £57k per year, launched the MRC. I mention the tuberculosis sanatoria because 37 years later Bradford Hill introduced the double blind randomised controlled trial design to clinical research, for a trial of streptomycin in the treatment of TB. The results, reflecting the design, were of such power and clarity that within a few years national death rates from TB had plummeted and the hundreds of sanatoria across the country, still open in the late 1940's, were closed by the mid 1950's. Within the same few years, Bradford Hill, together with Richard Doll, laid the foundations for modern chronic disease epidemiology in their studies of lung cancer. An initial case control study was followed by the prospective study of British doctors. The latter, with its quantification of the full health risks of smoking and of the benefits of quitting, has perhaps benefited human health across the world more than any other study in the last few decades. It is scandalous that the committee responsible did not see fit to honour the investigators with the Nobel Prize. Time moves on, risk for disease presents an increasingly complex problem and technology has developed explosively. When I became director of the unit in 1986 the principal question was what should be the role of the unit in a rapidly changing biomedical environment. The response has emerged over the years with great clarity. Advances in statistical theory over the past 30 years with concommitant developments in computing power have revolutionised the study of complex longitudinal disease processes. Together of course with the explosive advances in genetics. In the late 1980's it became clear that, with the unit in the vanguard, the development of Bayesian methods of analysis provided new and powerful approaches to the study of a wide range of diseases and potential interventions. The AIDS/HIV epidemic, cognitive decline and dementia, screening for cancer and the behaviour of precancerous lesions, were among the first areas where fresh statistical thinking provided basic new insights. Cambridge has provided a very fertile biomedical research environment for these initiatives to develop. With a new director, and a set of new research proposals, the unit is well placed to launch itself on the second hundred years of achievement.

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

Professor Vernon Farewell, MRC Biostatistics Unit

Prior to moving to the MRC Biostatistics Unit in 2001, Vern Farewell held professorial positions at the University of Washington, the University of Waterloo and University College London. He has published widely in the statistical and medical literature and is co-author of the four editions of the book "Using and Understanding Medical Statistics". Since 2007, he has been Editor of "Statistics in Medicine". His current research has a primary focus on the analysis of observational longitudinal data to address multiple questions related to the course of particular diseases. Particular aspects of this work are the development of analysis methods that account for the correlation of observation from a single subject over time, often through random effects models, and the use of multi-state structures to appropriately model change in disease over time. His group’s applied collaborative medical research includes long-standing collaborations in studies of rheumatic diseases, particularly psoriatic arthritis and systemic lupus erythematosus, but also covers a spectrum of other diseases.

Seminar Title: Historical perspective: Dr. John Brownlee, First 'Director' of the MRC Statistical Department Monday 24 March, 11.30 – 13.00 session

Abstract: In July 1914, Dr. John Brownlee was appointed head of the Statistical Department of the newly formed Medical Research Committee. He had qualified in mathematics, natural philosophy, and medicine at the University of Glasgow, and by 1914 had established a reputation as a public health officer, an expert in infectious diseases, and a proponent of Pearsonian statistics applied to medicine. This presentation will examine his professional career and his character.

Professor Peter J Diggle, and

Peter Diggle is Professor in the Department of Epidemiology and Population Health, University of Liverpool, and Distinguished University Professor of Statistics in the Faculty of Health and Medicine, Lancaster University. He also holds Adjunct positions in Johns Hopkins University School of Public Health, the International Research Institute for Climate and Society, , and School of Public Health. Peter's research involves the development of statistical methods for spatial and longitudinal data analysis, and their applications in the biomedical and health sciences. His current projects include a study of the spatial epidemiology of campylobacter in the UK, forecasting emergent meningitis epidemics in sub-Saharan Africa and the detection of incipient renal failure in primary care patients. He is also a member of the North of England Health e-Research Centre, an MRC-funded consortium of researchers in Manchester, Lancaster, Liverpool and York, where he is developing statistical methods for real-time analysis of routinely recorded health information.

Seminar Title: Informative missingness and/or informative non-missingness: what can spatial epidemiology learn from longitudinal data analysis? Monday 24 March, 14.00-16.30 session

Abstract: Those involved in the design and/or analysis of longitudinal studies are usually well-acquainted with the need to think carefully about issues around potentially informative missing values, on which topic there is an enormous literature. A smaller literature considers the converse problem of accommodating potentially informative non- missingness. This arises when study participants are measured when they make themselves available, rather than at pre-specified follow-up times. In spatial epidemiology, especially in low-resource settings, it is rare that either of these issues is treated seriously: spatial sampling schemes are typically opportunistic. But there are good practical reasons, again especially in resource settings, why opportunistic sampling is the only viable option. This presents an interesting challenge to the use of model-based statistical methods for analysing the resulting data.

Professor Arnoldo Frigessi, Department of Biostatistics, University of Oslo

Arnoldo Frigessi is Professor of Statistics at the Department of Biostatistics at the University of Oslo and director of Statistics for Innovation, a centre of excellence for research-based innovation. His research interests include highly structured stochastic models, computationally intensive inference, statistical genomics, multivariate tail dependence, with applications in molecular biology, infectious diseases, insurance and climate research.

Seminar Title: Integrative genomics: learning from layers of data Wednesday 26 March, 13.30-15.00 session

Abstract: Recent and emerging -omics technologies enable measurement of molecular phenotypes at unprecedented levels of breadth and resolution. We can now view a specific molecular biological system from many perspectives, producing various type of data that capture different phases of the biological dynamics. But despite important successes, the fundamental complex dependences between various molecular regulations are often not fully understood. A deeper understanding would help to make more precise predictions of disease progression and therapeutical efficacy. A major outstanding aim is the development of methods that allow the joint analysis of multiple -omics components. In this lecture I will first give a systematic review of recent statistical methodology that exploits collectively multivariate -omics data sets. Then I will report on our own Bayesian model based approach for the unsupervised classification of breast cancer patients based on multiple layers of -omics data.

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

Professor Robin Henderson, School of Mathematics and Statistics, University of Newcastle

Rob Henderson is Professor of Statistics and Head of Mathematics and Statistics at . His expertise relates to biostatistical applications, including event history, dynamic treatment, repeated measurement and missing data methods. Research Interests: Biostatistics; Longitudinal data analysis; Survival and event history analysis; Missing data; Optimal dynamic treatment regimes.

Seminar Title: Dynamic analysis of longitudinal and event history data Wednesday 26 March, 09.00-10.30 session

Abstract: Longitudinal and event history data accrue over time. Dynamic models explicitly build conditioning on the past into model structures. For example transition models in longitudinal studies or intensity models in event history would be classed as dynamic whereas random effects or rate models would not. We argue that dynamic models provide a natural approach for predictive purposes and we consider the issue of dynamically measuring predictive accuracy, especially in the presence of missing data. How much information is required before we become confident in short, medium and long-term predictions? We illustrate the ideas through the analysis of three longitudinal studies of incidence and prevalence of infant diarrhoea in Salvador, Brazil, carried out in the same areas but in different years.

Professor Andy Grieve, Clinical Research Methodology, Aptiv Solutions

Andy Grieve is Senior Vice President for Clinical Trials Methodology at Aptiv Solutions based in Stevenage, UK. From 2006 to 2010 he was Professor of Medical Statistics at King’s College London. Prior to joining King’s he spent over 30 years in the pharmaceutical industry working for CIBA-GEIGY, ICI Pharmaceuticals (Zeneca) and Pfizer. Andy is a Fellow, Chartered Statistician and former president of the Royal Statistical Society; Fellow of the American Statistical Association and honorary life-member of Statisticians in the Pharmaceutical Industry of which he is a past-Chairman and founder-member. His research has been primarily concerned with the application of statistics to the pharmaceutical industry, and in particular he has concentrated on the implementation of Bayesian ideas and techniques. Latterly his research has concentrated on the design and implementation of Bayesian adaptive trials. Andy has published over 120 articles and is the author of a book for non-statisticians involved in clinical trials: FAQ's on Statistics in Clinical Trials. Andy received his MSc in Statistics from Southampton University in the UK in 1975, a Ph.D. in Statistics from Nottingham University in the UK in 1992 and an Honorary Doctorate from Kingston University for Services to Statistics in 2006.

Seminar Title: Adapting for Success: The Genesis of Adaptive Designs Wednesday 26 March, 11.00-12.30 session

Abstract: The last decade and a half has seen considerable interest in adaptive designs in medical and pharmaceutical research. One of the perceived barriers to their widespread use is their novelty. In this talk I look at the genesis of adaptive designs and show that far from being new they have been available, though under-utilised, for 80 years.

Professor Peter Holmans, MRC Centre for Neuropsychiatric Genetics & Genomics, Cardiff University

Peter Holmans is Professor of Biostatistics & Genetic Epidemiology in the Biostatistics & Bioinformatics Unit (BBU) in theWales College of Medicine, Cardiff University, UK, and direct the research in statistical genetics carried out by the BBU. The BBU has close links with several departments carrying out large-scale linkage and association studies of complex traits. These include studies of schizophrenia, bipolar disorder, late-onset Alzheimer's disease, ADHD and dyslexia currently being carried out in the Department of Psychological Medicine. The BBU works with numerous international collaborators, notably as part of the Psychiatric Genetics Consortia and the International Genomics of Alzheimers Project. Over the past 5 years, BBU members have been authors on over 150 peer-reviewed papers, and named applicants on awarded grants totalling over £8 million. Peter has a long-standing interest in the analysis of genome-wide linkage and association studies of complex genetic traits. He has also recently become involved in the analysis of gene expression data, and linking this to large genomic datasets to find genetic variation relevant to disease. He has taken an active role in developing novel statistical methodology for linkage and association analysis of complex genetic traits, notably in the use of covariates in linkage and association studies, and the effects of genotyping error on genetic studies. Currently, he is particularly interested in the analysis of functional pathways in genome- wide association, CNV, gene expression and next-gen sequencing data, and combining evidence for disease susceptibility across different types of genetic variation.

Seminar Title: Extracting biological meaning from genomic data using pathway analysis: application to schizophrenia Wednesday 26 March, 16.00-17.30 session

Abstract: Schizophrenia is a debilitating psychiatric disorder, with a prevalence of around 1% in the general population. As such, it causes a considerable public health burden. Twin studies have shown a high genetic component (heritability) to schizophrenia susceptibility. Recently, large genome-wide association studies have found several convincing associations between common variants and schizophrenia. However, these do not explain all the heritability, thereby implicating rare variation as important in the aetiology of schizophrenia. Power to detect association to a single variant or gene is limited due to the rarity of the variants , but can be increased by combining biologically-related genes (“pathways”) into a single analysis. Pathway analysis results of rare copy number variants (CNVs) and denovo single nucleotide variants (SNVs) are presented. Both types of variation show convergent evidence for enrichment of association signal to synaptic pathways. Overlap in signal from other neurodevelopmental disorders to these pathways will also be presented.

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

Professor Chris Holmes, Department of Statistics, Chris Holmes is Professor of Biostatistics and Fellow of Lincoln College. He moved to Oxford from Imperial College London in February 2004. At Imperial College he studied for his doctorate in Bayesian statistics, investigating novel nonlinear pattern recognition methods. This was followed by a post-doctoral position and then a lectureship at Imperial. Previous to this Chris worked in industry for a number of years researching in scientific computing, developing techniques for real-time pattern recognition models in defence and SCADA (Supervisory Control and Data Acquisition) systems. His current research is focussed on applications and statistical methods development in the genomic sciences and genetic epidemiology. He holds a programme leaders grant in Statistical Genomics from the Medical Research Council.

Seminar Title: Computational decision theory and Bayesian methods in genomics Tuesday 25 March, 15.30-17.00 session

Abstract: Statistical analysis in genomics is complicated by the typical high-dimensional linear noisy data structures that are produced from modern high-throughput assays. Examples include the analysis of cancer genomes as well as population case-control genome-wide association studies. In this talk I will discuss how computational decision theory and Bayesian methods can be used to help draw robust conclusions from such experiments.

Professor Dirk Husmeier, School of Mathematics and Statistics, University of Glasgow He currently holds a Chair of Statistics at Glasgow University, which he took up in October 2011 after previous employment at Biomathematics & Statistics Scotland (formally part of the James Hutton Institute), 1999-2011, and Imperial College London, 1997-1999. Coming from a background in theoretical physics and applied mathematics, his research focuses on the development of novel statistical and machine learning methods for bioinformatics and computational biology, with an emphasis on Bayesian inference. His recent research projects were related to molecular phylogenetics, pattern recognition in DNA sequence alignments, the detection of intraspecific recombination in bacteria and viruses, the reconstruction of gene regulatory networks from transcriptomic profiles and postgenomic data integration, and the development of improved MCMC samplers for Bayesian learning of Bayesian networks. His current research focuses on improved Bayesian hierarchical models for the prediction of molecular regulatory networks subject to adaptation, the inference of species interaction networks in ecology, and Bayesian inference in mechanistic models of molecular pathways.

Seminar Title: Bayesian modelling and inference in complex biosystems Tuesday 25 March, 15.30-17.00 session

Abstract: The topic of this talk will be related to inference of gene regulatory networks from postgenomic data. The speaker will briefly compare approaches based on statistics versus mechanistic models, and discuss a Bayesian hierarchical modelling framework. Various applications to synthetic biology of Saccharomyces cervisiae, circadian regulation in Arabidopsis thaliana, and simulated data using Markov jump processes will be discussed. Another topic of discussion will be parameter inference in mechanistic models of biochemical pathways with nonparametric Bayesian statistics based on Gaussian processes.

Professor Sharon J. Hutchinson, Glasgow Caledonian University and Health Protection Scotland

Sharon Hutchinson is a Professor of Epidemiology and Population Health at Glasgow Caledonian University, and works in association with Health Protection Scotland to develop surveillance on blood-borne viruses. She manages a broad translational research programme, which is both methodological and applied and spans activities including infectious disease modelling, record-linkage studies and evaluation of public health interventions. Her research provided the key evidence to guide a public health response to Scotland’s hepatitis C epidemic, which culminated in the Scottish Government investing significantly in improving services. She is currently supporting WHO on initiatives to tackle viral hepatitis in w lo and middle income countries. In 2009, in recognition of the major contribution to the science and practice of public health, Sharon was awarded with Honorary Membership to the Faculty of Public Health of the Royal Colleges of Physicians in the UK. In 2013, she became a member of the Royal Society of Edinburgh’s Young Academy of Scotland.

Seminar Title: Translating research into public health policy: Scotland’s Hepatitis C Odyssey Monday 24 March, 16.00-17.30 session

Abstract: In 2004, Scotland’s Health Minister recognised that the hepatitis C virus (HCV) was one of the country’s most challenging public health concerns. The Scottish Government thereafter launched an Action Plan on HCV, which aimed to: (i) prevent the spread of infection, particularly among people who inject drugs (PWID), (ii) diagnose HCV-infected people, and (iii) ensure that those infected receive optimal treatment, care and support. The Plan was a two-phased one. Phase I involved gathering evidence to inform proposals for the development of HCV services during Phase II. Phase II, launched in May 2008, saw serious commitment to tackle the HCV challenge facing Scotland, with an investment of approximately £43 million over three years, to deliver actions designed to improve prevention, diagnosis and treatment services. Scotland’s HCV Action Plan is regarded as a model of good practice. It was based on an extensive evidence base and consultation process, adopted a multidisciplinary approach and was performance managed. The focus here will be on the approaches adopted to generate the key evidence and to assess performance, including: anonymous blood-spot testing surveys to monitor HCV incidence and prevalence, case-finding evaluations, record-linkage exercises, and modelling studies to forecast the burden of HCV disease.

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 Christopher H. Jackson, MRC Biostatistics Unit Chris Jackson is a senior statistician at the MRC Biostatistics Unit, working on statistical methods for evidence synthesis in health economic evaluations and public health policy. The models used in these areas are subject to many uncertainties in assumptions, choices of data, and extrapolation of results over time. Statistical methods must ensure that the decisions based on these models accurately reflect the available evidence and the extent of uncertainty. Methods for model comparison, flexible modelling and model calibration are used, typically within a Bayesian framework. Chris's research interests also include multi-state models for longitudinal data, particularly data arising from disease progression. He maintains the "msm" R package for continuous-time Markov and hidden Markov modelling, several other R packages, and contributes to the BUGS software for Bayesian analysis.

Seminar Title: Multi-state modelling software and encouraging statistical software development Tuesday 25 March, 13.30-15.00 session

Abstract: Firstly I will discuss software for multi-state modelling in continuous time. I will describe the design principles of my own R package "msm" and its current capabilities and limitations. I will then mention some possible ways the package might be extended in the future. While msm was designed for intermittently-observed multi-state processes, I will also discuss other software more suitable for fully-observed processes, such as the "survival" and "mstate" packages for R. Secondly I will discuss statistical software development more generally. For new methods to become usable, they need accessible software. I will give some thoughts about how to encourage more software development in the biostatistics research community, and open the subject for discussion.

Dr Anthony L. Johnson, MRC Biostatistics Unit andMRC Clinical Trials Unit at UCL

Tony Johnson Ph.D., CStat, is Past Programme Leader at MRC Biostatistics Unit, Cambridge and Senior Medical Statistician at MRC Clinical Trials Unit at UCL, London.

Seminar Title: Historical perspectives- Monday 24 March, 11.30-13.00 session

Abstract: The Second Director, Major Greenwood, (1880 – 1949) Tony Johnson and Vern Farewell Major Greenwood was the foremost medical statistician of the first half of the 20th century in the UK. He founded the first departments of medical statistics and gave the first courses in the subject. From around 1920 he worked side-by- side with the Unit’s first Director, John Brownlee, to oversee the statistical work of the MRC. In 1927 he became the first Professor of Epidemiology and Vital Statistics at the London School of Hygiene and Tropical Medicine and on Brownlee’s death in the same year he also became the Unit’s second Director. Between them they governed MRC’s Statistical Department for the first thirty years and established its character, its remit, and its durability that lead ultimately to the appointment of six further eminent Directors starting with in 1946 and taking it to the present day. Greenwood’s career from 1898 onwards will be described with his achievements and some debate about two important issues that he did not tackle.

Dr David Lunn, MRC Biostatistics Unit

Dave Lunn is a senior statistician at MRC Biostatistics Unit and a co-author of the BUGS software. His research interests include: dynamical systems; genetic epidemiology; infectious disease dynamics; model uncertainty; pharmacokinetics/pharmacodynamics; glucose-insulin kinetics/dynamics; reversible jump MCMC; cutting feedback in graphical models.

Seminar Title: Historical perspectives Monday 24 March, 11.30-13.00 session

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

Professor Alexander McNeil, Department of Actuarial Mathematics and Statistics, Heriot-Watt University, Edinburgh

Alexander McNeil is Maxwell Professor of Mathematics in the Department of Actuarial Mathematics and Statistics at Heriot-Watt University. He is also Director of the Scottish Financial Risk Academy (SFRA), which organises knowledge exchange activities between the university and financial sectors in Scotland including Risk Colloquia, training events and postgraduate placements in industry. Formerly Assistant Professor in the Department of Mathematics at ETH Zurich he has a BSc in mathematics from Imperial College, London and a PhD in mathematical statistics from Cambridge University. His interests lie in the development of mathematical and statistical methodology for integrated financial risk management and include extreme value theory (EVT), risk theory, financial time series analysis and the modelling of correlated risks. He has published papers in leading statistics, econometrics, finance and insurance mathematics journals and is a regular speaker at international risk management conferences. He is joint author, together with Rüdiger Frey and Paul Embrechts, of the book "Quantitative Risk Management: Concepts, Techniques and Tools", published by Princeton University Press in 2005.

Seminar Title: Copulas: From Biostatistics to the Financial Crisis Monday 24 March, 16.00-17.30 session

Abstract: Professor McNeil will give a short tour through some of his work on copulas and dependence models and explain how this relates to shared frailty models in survival analysis and the work of researchers at the MRC Biostatistics Unit. He will also talk about the use of copula models in financial risk modelling and their notorious role in the 2007-9 financial crisis.

Dr Richard Nixon, Advanced Qualitative Sciences, Novartis Pharma AG, Basel, Switzerland

Richard is a Senior Expert Modeler, in the Advanced Quantitative Sciences group in Novartis, Basel. His areas of interest are applying Decision Analysis methods to bridge between clinical and commercial decisions; Benefit-Risk methods; Health Economic modeling; Evidence synthesis and Bayesian Methods. He holds a Degree in Mathematics from the University of Durham, a Diploma in Mathematical Statistics from the University of Cambridge and a Ph.D. in Biostatistics from the MRC Biostatistics Unit in Cambridge.

Seminar Title: Why the MRC-BSU is a great place to learn your trade Monday 24 March, 16.00-17.30 session

Abstract: A key role for an applied statistician is to be a translational scientist capable of understanding a problem, translating it to a model, and communicating back the results. This role requires a portfolio of skills. The statistician needs to have an understanding of the scientific subject area, a solid understanding of statistics and good communication skills. Our time at the MRC-BSU grounded us in all these areas. The BSU sits at the sweet spot between methodology and application. Real world application drives the development of the methodology, and the methodology is grounded in solid theory. Much of the skills of being a translational modeller are tacit, a text book cannot teach the most valuable lessons from you gain from the BSU. Working with seasoned experts, learning how they approach a problem and the process they use to solve it, was a great way to gain these key skills. In this talk we will also present case studies in the application of statistics and decision analysis to support drug development.

Dr David Ohlssen, Novartis Pharmaceuticals Corporation

David Ohlssen is currently a senior expert statistical methodologist and Bayesian team lead, within the Novartis statistical methodology group, based in East Hanover New Jersey. Since joining Novartis in 2007, he has developed a broad range of experience in applying novel statistical approaches within a drug development setting. Previously, after completing his PhD in Biostatistics at the University of Cambridge, he worked as a research fellow at theMRC Biostatistics Unit (Cambridge UK), where his interests included: diagnostics for Bayesian models, novel clinical trial design and statistical methods for the profiling of health-care providers. His professional activities include serving as a member of the recently-formed Bayesian DIA Working Group and within the group acting as the chair of the safety meta-analysis sub-team.

Seminar Title: Bayesian approaches to assessing treatment effect heterogeneity in drug development Monday 24 March, 16.00-17.30 session

Abstract: The topic of treatment effect heterogeneity across sub-populations is of critical importance in drug development and comparative effectiveness. There have been a number of recent reports and guidance documents related to this issue: For example: The FDA draft Guidance for Industry on Enrichment Strategies for Clinical Trials and the recent report to the PCORI Methodology Committee on standards in addressing heterogeneity of treatment effectiveness. In this talk, we explore a general modelling strategy for exploring treatment effect heterogeneity across sub-populations. Various options for Bayesian shrinkage estimation are examined and then applied to a case-study, in which evidence was available from seven phase II-III clinical trials.

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

Professor Mahesh Parmar, MRC Clinical Trials Unit and University College London

Mahesh Parmar is Director of the MRC Clinical Trials Unit and Professor of Medical Statistics and Epidemiology at University College London (UCL). He is also Director of the newly formed Institute of Clinical Trials and Methodology at UCL. He has been an Associate Director of the National Cancer Research Network since its inception in 2001, an organisation which has more than trebled the number of patients going into cancer studies in England. Max joined the MRC in 1987. He has more than 200 publications in peer reviewed journals, many of which have had direct impact on policy, clinical practice and improving outcomes for patients. The Unit he directs is at the forefront of resolving internationally important questions, particularly in infectious diseases and cancer, and also aims to deliver swifter and more effective translation of scientific research into patient benefits. It does this by carrying out challenging and innovative studies and by developing and implementing methodological advances in study design, conduct and analysis.

Seminar Title: Randomised Trials: New Designs for New Challenges Wednesday 26 March, 11.30-13.00 session

Abstract: There are a number of key challenges in clinical trials. There is a pressing need to evaluate new therapies more quickly. There is also a need to improve our success rate at identifying new effective therapies. Alongside these challenges we need to find a way to efficiently evaluate the many new targeted therapies in the pipeline, together with their putative biomarkers which may help identify groups of patients who benefit most/least from these new therapies. In this presentation we present new designs to deal with these challenges. These new designs have been deployed in large-scale clinical trials which will also be presented to show how they might be deployed and implemented.

Professor Adrian E. Raftery, Center for Statistics and the Social Sciences, Center for Studies in Demography and Ecology, University of Washington

Adrian Raftery is Blumstein-Jordan Professor of Statistics and Sociology and a faculty affiliate of the Center for Statistics and the Social Sciences and the Center for Studies in Demography and Ecology at the University of Washington. He works on the development of new statistical methods for the social, environmental and health sciences. Adrian is currently on leave from UW and is serving as E.T.S. Walton Visiting Professor, School of Mathematical Sciences at the University College Dublin, Ireland. He studied Mathematics and Statistics at Trinity College Dublin, Ireland, and obtained his Doctorate in Mathematical Statistics in 1980 from the Université Pierre et Marie Curie in Paris, France, advised by Paul Deheuvels. From 1980 to 1986 he was a Lecturer in Statistics at Trinity College Dublin, and since then he has been on the faculty of the University of Washington, and was founding director of the university's Center for Statistics and Social Sciences.

A member of the National Academy of Sciences, Raftery is a fellow of the American Statistical Association, Institute of Mathematical Statistics and the American Academy of Arts and Sciences. Among his significant honours is the 2011 ASA Award for Outstanding Statistical Application, the 2011 ASA Statistics in Chemistry Award, the Jerome Sacks Award for Outstanding Cross-Disciplinary Research for the National Institute of Statistical Sciences as well as the H.O. Hartley Memorial Lecturer at Texas A&M University, the 2012 Emanuel and Carol Parzen Prize for Statistical Innovation. Adrian was elected to the Royal Irish Academy in 2013.

Seminar Title: Bayesian Reconstruction of PastPopulations for Developing and Developed Countries Tuesday 25 March, 11.00-12.30 session

Abstract: I will describe Bayesian population reconstruction, a new method for estimating past populations by age and sex, with fully probabilistic statements of uncertainty. It simultaneously estimates age-specific population counts, vital rates and net migration from fragmentary data while formally accounting for measurement error. As inputs, it takes initial bias-corrected estimates of age-specific population counts, vital rates and net migration. The output is a joint posterior probability distribution which yields fully probabilistic interval estimates of past vital rates and population numbers by age and sex. It is designed for the kind of data commonly collected in demographic surveys and censuses and can be applied to countries with widely varying levels of data quality. This is joint work with Mark Wheldon, Patrick Gerland and Samuel Clark.

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

Professor , MRC Biostatistics Unit and University of Cambridge

Sylvia Richardson is the Director of the MRC Biostatistics Unit and holds a Research Professorship in the University of Cambridge since 2012. Prior to this Sylvia held the Chair of Biostatistics in the Department of Epidemiology and Biostatistics at Imperial College London since 2000 and was formerly Directeur de Recherches at the French National Institute for Medical Research INSERM, where she held research positions for 20 years. She has worked extensively in many areas of biostatistics research and made important contributions to the statistical modelling of complex biomedical data, in particular from a Bayesian perspective. Her work has contributed to progress in epidemiological understanding and has covered spatial modelling and disease mapping, measurement error problems, mixture and clustering models as well as integrative analysis of observational data from different sources. Her recent research has focussed on modelling and analysis of large data problems such as those arising in genomics. Sylvia has a Doctorat Es Sciences from the University of Paris XI, a PhD in Probability Theory from the , and had held lectureship positions at Warwick University and the University of Paris V. Sylvia is currently involved in several collaborative projects in genomics. She has a broad range of external activities, notably on scientific boards with ANSES (Paris) and the Centre of "Statistics for Innovation" (Oslo). In 2009, Sylvia was awarded the Guy Medal in Silver from the Royal Statistical Society and a Royal Society Wolfson Research Merit award.

Seminar Title: Invited closing session Wednesday 26 March, 13.30-15.00 session

Professor Simon Thompson, Department of Public Health and Primary Care, University of Cambridge

Simon Thompson is Director of Research in Biostatistics in the Cardiovascular Epidemiology Unit within the Department of Public Health and Primary Care, University of Cambridge, and is a Fellow of the Academy of Medical Sciences. He was Director of the MRC Biostatistics Unit from 2000-2011. He held previous academic appointments at the London School of Hygiene and Tropical Medicine, and as the first Professor of Medical Statistics and Epidemiology at Imperial College London.

His research interests are in meta-analysis and evidence synthesis, clinical trial methodology, health economic evaluation, and cardiovascular epidemiology; he has published widely in these areas. He has collaborated on a number of major clinical trials, recently including all the major UK national trials of screening and treatment for abdominal aortic aneurysms. He has been a strong advocate for biostatistics through his research papers, didactic articles, and contributions to courses and workshops both in the UK and abroad. He has also taken on a number of responsibilities for the MRC, the Royal Statistical Society, and other international professional societies.

Seminar Title: Biostatistics in the st21 Century? Monday 24 March, 14.00-15.30 session

Abstract: I will give a personal view on some issues facing biostatistics, as a subject and as a profession. Coloured by my time at BSU, I will reflect on research training, revelations from job interviews, the publication process, and the emergence of large collaborative research networks.

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

Duncan C. Thomas, Biostatistics Division and Department of Preventive Medicine, University of Southern California

Duncan Thomas is Professor of Preventive Medicine, Director of the Biostatistics Division, and Verna R. Richter Chair in Cancer Research at the University of Southern California Keck School of Medicine. He received his undergraduate degree from Haverford College, an M.S. in Mathematics from Stanford University, and a Ph.D. in Epidemiology and Biostatistics from McGill University in 1976. His primary research interest has been in the development of statistical methods for cancer epidemiology, but he also has wide ranging interests in both environmental and genetic epidemiology. His statistical contributions include methods for analysis of nested case-control studies, approaches to modelling exposure-time-response relationships and interaction effects, exposure modelling and measurement error, and the use of Markov chain Monte Carlo (MCMC) methods and Generalized Estimating Equations (GEE) methods in genetics. On the environmental side, he has been particularly active in radiation carcinogenesis, having collaborated on studies of cancer in residents downwind of the Nevada Test Site, uranium miners, medical irradiation, and the atomic bomb survivors. He was a member of President Clinton’s Advisory Committee on Human Radiation Experiments, as well as the National Academy of Sciences Committee on the Biological Effects of Ionizing Radiation (BEIR V), and radiation advisory committees for numerous other governmental agencies. Other environmental activities include studies of asbestos, malathion spraying in California, electromagnetic fields, and air pollution; he is Co-Director of the Southern California Environmental Research Center. On the genetic side, Dr. Thomas has numerous publications in the area of statistical genetics and is collaborating on family studies of breast, ovarian, colon, prostate and other cancers, insulin dependent diabetes, systemic lupus erethematosis, and other diseases. He chairs organizing committees for the Genetic Analysis Workshop and the Informatics Consortium for the NCI Cooperative Family Registries for Breast and Colorectal Cancer, and is currently Past President of the International eneticG Epidemiology Society. These three broad areas of interest make him uniquely qualified to address methodological challenges in studying gene-environment interactions.

Seminar Title: Emerging challenges in statistical genetics Wednesday 26 March, 13.30-15.00 session

Abstract: Statistics is being profoundly challenged in the “big data” era, perhaps nowhere more dramatically in than in the genomics area. Here, “big” means particularly ultra-high dimensional data—the “large p small n” problem— although the proliferation of consortia means n is also often extremely large. The latter can pose computational and data storage challenges, but tends not to introduce fundamentally new conceptual problems. Statisticians have responded to the high dimensionality problem with a variety of approaches—Bayesian and frequentist—such as sparse penalized regression and hierarchical modeling. We begin by discussing these approaches in the genetics context, with examples of their use in genomewide association studies (GWAS) drawn from our own experience and that of investigators at the MRC Biostatistics Unit. As the genetics field moves into the “post-GWAS” era, many new challenges are arising. These include the design of next-generation sequencing studies and analysis of rare variant associations, inference on complex biological networks, integration of data and knowledge across different “-omics” platforms, the need for efficient computational approaches using parallel processing techniques, and the implications of all this for personalized medicine. Various emerging technologies, such as brain imaging and microbiome assays, can be expected to pose “big data” challenges that dwarf anything we’ve confronted yet.

Join us on twitter @MRC_BSU