Palmgren Symposium 2016
Statistical science in biomedical research: A look to the future
September 1-2, 2016
Nobel Forum, Karolinska Institutet, Stockholm
This two-day symposium focuses on statistical science in biomedical research. It ties together Juni Palmgren’s work and visions during more than three decades. When looking to the future, the modern features of data and computational sciences (Big Data, eScience and eInfrastructure) will increasingly label the field and will likely shape also the context and future of medical statistics.
The aim of the symposium is twofold:
- To articulate the scientific, political, educational and organisational challenges that lie ahead in order for data integration, modelling and simulation to penetrate the research agenda for complex problems in medical research. On this note international special invited speakers and a panel of local experts will feature.
- To get a glimpse of Palmgren’s former students, postdocs and protégés from Helsinki and Stockholm, who will give a personal view of their careers and their hopes for the future.
Biography: Juni Palmgren
Juni Palmgren started her career at the University of Helsinki and at the National Public Health Institute in Finland. She was recruited as Professor of Biostatistics to the Department of Mathematical Statistics at Stockholm University 1997. She held a Guest Professorship at Karolinska Institutet from 1999 and was awarded an Honorary Doctorate in Medicine MDHc 2007 for her efforts to promote Biostatistics at Karolinska Institutet. She was Secretary General for Research Infrastructure at the Swedish Research Council 2010-2015 and Finland Distinguished Professor, FiDiPro at the Institute for Molecular Medicine Finland, FIMM 2011-2015. In 2013 she was appointed full Professor of Biostatistics at Karolinska Institutet. Her academic home is the Department of Medical Epidemiology and Biostatistics.
Register here: Registration will close when all seats are taken! Please note that there are 100 seats available.
Thursday September 1, 2016
|13.00-13.20||Opening/Juni Palmgren as mentor and visionary|
|Analytic methods for large and complex data|
|13.20-14.00||Arnoldo Frigessi: Just data is not enough: the Oslo experience in model based statistics for big health data.|
|14.00-14.20||Samuli Ripatti: Large-scale studies in complex disease genetics|
|14.20-14.40||Olivia Eriksson: From Complex Dynamical Intracellular models to e-Science|
|14.40-15.00||Alexandra Jauhiainen: Working with different types of complex and large data - experiences from both academia and industry|
Panel discussion on Big Data, eScience and eInfrastructure in biomedical research
Participants: Arnoldo Frigessi (Oslo), Els Goetghebeur (Gent), Henrik Grönberg (KI), Erik Lindahl (KTH), Martin Ingvar (KI), Olli Kallioniemi (SciLifeLab), Louise Ryan (Sydney)
Moderator: Hans-Olov Adami
|18.00-||Dinner (please note: registration required to attend!)|
Friday September 2, 2016
|Complex statistical models in education and research|
|09.00-09.40||Els Goetghebeur: Cost effective models for prediction and causal effect estimation: the goldmine of disease registers|
|09.40-10.00||Arvid Sjölander: Causal inference in theory and practice|
|10.00-10.20||Jouni Kuha: Statistics in the social sciences: Themes and developments|
|10.20-10.40||Monica Leu: From population-based multigeneration data to nutritional research|
|Statistical science serving industry and the society|
|11.10-11.30||Annica Dominicus: In a world of statistical models – from twins to drug development|
|11.30-11.50||Pasi Korhonen: Use of Nordic nationwide databases in evaluation of drug safety issues and Causal models applied in clinical trials|
|11.50-12.10||Gudrun Jonasdottir: Solving the problem - reflections on how a Ph.D. in Biostatistics have impacted my career|
|The necessity and relevance of statistical science|
|13.00-13.20||Patricia Geli: Stories from the frontiers of development|
|13.20-13.45||Louise Ryan: Clever subsampling can save time and boost power!|
Special invited speakers
Arnoldo Frigessi, University of Oslo
Arnoldo Frigessi is professor of statistics at the University of Oslo and director of BigInsight, a centre for research-based innovation, a consortium of industry, business, public actors and academia, producing research results from big data in various sectors of the Norwegian knowledge based economy. He has a partial affiliation also with the Oslo University Hospital. Frigessi leads the Oslo Center for Biostatistics and Epidemiology. His research interests include stochastic models, computationally intensive statistics, big data, preference learning, multivariate dependence, with applications to molecular biology, sensor data, cancer, personalized marketing.
Title: Just data is not enough: the Oslo experience in model based statistics for big health data.
Abstract: BigInsight is a Norwegian center for research based innovation, that develops statistical and machine learning methods to analyse complex and huge data. We focus on two main themes: personalise solutions and forecasting transient dynamics. I will give an overview of some of our projects in the area of big data in health and argue that statistical models and methods are often needed to harvest value from big data. Integrative genomics, pan cancer studies, mathematical models of treatment in cancer and patient safety will be examples. I will conclude with some thoughts on the future of the big health data era.
Els Goetghebeur, Ghent University
Els Goetghebeuris Professor of Statistics at Ghent University, Belgium. She got her graduate training in mathematics (KUL) and Statistics (LUC) and held faculty positions at the London School of Hygiene and Tropical Medicine (UK) and Maastricht University (NL). She taught at the Harvard School of Public Health and at Stanford University. Her research focuses on causal inference generally, on survival analysis and missing data problems, and more recently on big data problems in genetics (digital pCr and validation of NGS in the cancer clinic). She gets involved in several projects on quality of care in Belgium (hospitals), Flanders (Care centers) and Sweden (in collaboration with Umea University and Riks-stroke). She runs a consulting lab at Ghent University and is one of the founders of FLAMES, FLAnders training network in MEthodoloy and Statistics.
Title: Cost effective models for prediction and causal effect estimation: the goldmine of disease registers
Abstract: We use health register data to estimate causal effects and find that limited missingness in covariates can have a substantial impact. We therefore seek cost-effective predictors to adjust for confounders. An adapted lasso yields a model selection method that accounts for the cost of the entered covariates with a cross validation method that targets causal effects which are not directly observed. We apply the approach to estimate quality of care based on Riks-Stroke.
Louise Ryan, University of Technology Sydney
After completing her undergraduate degree in statistics and mathematics at Macquarie University, Louise Ryan left Australia in 1979 to pursue her PhD in statistics at Harvard University. After her postdoctoral training, she joined the faculty in the Department of Biostatistics at the Harvard School of Public Health, with a joint appointment at Dana-Farber Cancer Institute. She stayed there for 25 years, eventually becoming the Henry Pickering Walcott Professor and Chair of the Department. Louise returned to Australia in early 2009 as Chief of the Division of Mathematics, Informatics and Statistics in the Commonwealth Scientific and Industrial Research Organization (CSIRO). In 2012, she joined UTS as a distinguished professor of statistics in the School of Mathematical and Physical Sciences. Louise is well known for her contributions to the development of statistical methods for cancer, reproductive health and environmental health research. She loves the challenge and satisfaction of multi-disciplinary collaboration and is passionate about training the next generation of statistical scientists.
Title: Clever subsampling can save time and boost power!
Abstract: The classic case control study is a well known example of how clever sampling can boost study power, whilst managing study costs (in terms of numbers of subjects being studied). In this talk, we discuss several extensions to the classic case-control idea. In one example, oversampling cases and high exposed individuals can significantly boost the power of a study to assess gene-environment interactions. In a second example, the case-control concept is extended to a clustered data setting in a “big data” context. We show that with appropriate adjustment to reflect the sampling mechanism, case-control sampling can not only recover covariate effects of interest, but also lead to reliable estimates of cluster-specific random effects and variance components. We illustrate our findings with an analysis of rare adverse events in a population of blood donors drawn from multiple donation centres across Australia.
Former students, postdocs and protégés
Annica Dominicus, Scandinavian Development Services
Annica Dominicus defended her thesis in mathematical statistics in 2006 after which she joined AstraZeneca where she worked in different areas of drug development from early to late phase development. Since 2012 she has been working as a biostatistics consultant. In 2014 she joined Scandinavian Development Services where she works on a wide range of indications and statistical applications in medical research and drug development.
Title: In a world of statistical models – from twins to drug development
Abstract: Starting out in the field of twin modeling, a journey in a world of statistical models will be described. The role of statistics in drug development, and how it has changed over the last decade, will be discussed. Examples will include modeling and simulation in the planning of studies/programs, classification analyses and extreme value modeling of laboratory data.
Olivia Eriksson, Stockholm University
Olivia Eriksson received her PhD from Stockholm University in theoretical chemistry (computational biology) in 2008. After a postdoc position in Systems Biology at the Mathematical Statistics section of the Dept of Mathematics, Stockholm University she became coordinator and researcher at the Swedish e-Science Research Centre (SeRC), where she combines computational research on neurons in the brain with coordinating the different joint ventures of SeRC.
Title: From Complex Dynamical Intracellular models to e-Science
Abstract: Computational dynamical models describing intracellular phenomena are increasing in size and complexity as more information from experiments needs to be captured. I will describe different approaches to the non-trivial task of understanding how the outputs of such models depend on the model parameters and also how this led me in the direction of e-Science.
Patricia Geli, World Bank, Washington
Patricia Geli is currently Country Economist at the World Bank, Washington DC, USA. In this position, she is responsible for macroeconomic management, economic diversification, public investment management in fragile economies, in particular the Central African Republic (CAR). She is leading a large team preparing analytically rigorous Policy Notes that will form the basis of the World Bank's re-engagement with the CAR as the country transitions from fragility to resilience. She has had several missions in the World Bank since 2012 serving e.g. a number of countries in the Africa region. Notably, when the Ebola crisis hit, Patricia was dispatched to Sierra Leone on a long mission to work with the Government on the implementation of the Ebola Emergency Response Project—work for which she earned a special recognition. Prior to joining the World Bank, she was a post-doctoral researcher with Resources for the Future. In that capacity, she worked on innovative financing and assessed how tools from public economics could be applied to address public health problems, and led the research on a Gates-funded project that assessed the economic consequences of drug resistance and developed actionable policy proposals for low- and middle-income countries. In 2005, Patricia served as expert for the World Health Organization to evaluate surveillance systems of antimicrobial use and resistance in India and in South Africa. She has PhD in Mathematical Statistics from Stockholm University. In her spare time, Patricia serves as photographer for the National Press Club in Washington DC.
Title: Stories from the frontiers of development
Abstract: Post-conflict turmoil is prevalent in the developing world: 144 armed conflicts took place worldwide between 1989 and 2013. Common consequences include collapsed social service systems, pervasive infrastructure damage, widespread social distrust, death, destruction and loss of livelihoods. This presentation will explore the role of reliable data and, in particular, the Country Policy Institutional Assessments (CPIA). It will focus on lessons learned from two fragile states: Sierra Leone in the context of the Ebola outbreak, and the Central African Republic in the context of emerging from civil war.
Alexandra Jauhiainen, AstraZeneca
Alexandra Jauhiainen received her PhD in Mathematical Statistics from Chalmers in 2010. She has since worked as a postdoc both at the University of Michigan and at Karolinska Institutet. Since joining AstraZeneca, she has focused on statistical methods in early clinical development within the respiratory disease area. Her research interests include sparse modeling connected to uncertainty and calibration, surrogate endpoints and biomarkers in clinical development.
Title: Working with different types of complex and large data - experiences from both academia and industry
Abstract:Wide ranges of data fall under the term complex and/or large. The data can be made up of few observed units and many variables (sparse problem), observed as the output of a complex system, or dense with daily observations in thousands of patients. Research problems related to all these types of data will be illustrated while exemplifying how a research career can be shaped when working in both academia and industry.
Gudrun Jonasdottir Bergman, IMS Health
Gudrun Jonasdottir started her path to become a Biostatistician at the Department of Mathematics, Stockholm University. During her second year she had an option to study a variety of subjects, including statistics. In contrast to many of her fellow students, she found statistics to be fun and relatively easy. After finishing a Bachelor thesis at Statistics Sweden (SCB) she decided, upon advice from Juni Palmgren, to go for a M.Sc. in Applied statistics at Oxford. She returned to Sweden for PhD studies in Biostatistics at the Department of Medical Epidemiology and Biostatistics, and defended her thesis, “Genetic Association in the Presence of Linkage”, in 2008. She then turned outside academia with the first stop in the industry, at 4Pharma AB, working on regulations governing clinical trials. After her maternity leave, in the autumn of 2010, she switched to the National Board of Health and Welfare (Socialstyrelsen) working with Swedish Health Registries. In 2015 she took a new step in her career and joined IMS health.
Title: Solving the problem - reflections on how a Ph.D. in Biostatistics have impacted my career
Abstract: A relevant question to any student in statistics, considering becoming a Ph.D. student, is whether a Ph.D. degree has any impact on their future career. Studying and working to earn a Ph.D. degree is time consuming and hard work. Some may argue that the time is better spent by gaining experience from a work place. Others will argue that a Ph.D. trains you to become a better and more diverse problem solver. I will share experiences from my career and specifically how my time as a Ph.D. student has affected my career.
Pasi Korhonen, StatFinn and EPID Research
Pasi Korhonen did his PhD studies in biostatistics under Juni’s supervision 1995-2000. Pasi is an entrepreneur with 25 years’ experience in pharmaceutical statistics. Currently he leads two methodologically oriented companies. StatFinn is concentrated in statistics and data management in clinical trials and EPID Research focuses on evaluation of drug safety concerns by secondary use of existing data sources.
Title: Use of Nordic nationwide databases in evaluation of drug safety issues and Causal models applied in clinical trials
Abstract: Evaluation of bladder cancer risk with exposure to pioglitazone is presented as an example how the Nordic patient registers provide a useful platform for investigation of drug safety concerns. An application of causal modeling to correct for treatment crossovers is presented with a case study from oncology.
Jouni Kuha, London School of Economics and Political Science
Jouni Kuha studied Statistics at the University of Helsinki, where he obtained his MSc degree in 1992. He then received a PhD in Social Statistics from the University of Southampton in 1995, before working as a postdoctoral researcher at Nuffield College, Oxford, until 1999, and as an Assistant Professor of Statistics at the Pennsylvania State University in 1999-2001. Since 2001 he has been at the London School of Economics and Political Science, where he is now Associate Professor of Statistics and Research Methodology. His statistical research concerns various instances of the topics of categorical data analysis, models for imperfect or incomplete data, and latent variable modelling. The applied context of this work has shifted over time, from epidemiology to his current work on applications in the social sciences, such as the analysis of social mobility, trust in the criminal justice system, and predicting results of elections.
Title: Statistics in the social sciences: Themes and developments
Abstract: My own research career in statistics started with a problem in epidemiology: measurement error in self-reported nutrient intake, and how to allow for this error in estimating models for disease outcomes. Currently most of my work is motivated by questions in the social sciences and methods commonly used there, for example measurement of constructs in cross-national social surveys. However, although this shift in areas of application has been substantial, the journey in the space of statistical methods has been much shorter. This is because many methodological challenges, and current solutions to them, are shared between biomedical research – especially epidemiology and public health – and the social sciences. In this talk I discuss some of these similarities. I focus on two broad themes which are ubiquitous challenges in social research: measurement and causal inference from observational data.
Monica Leu, University of Gothenburg
Monica Leu has an undergraduate training in mathematical statistics and computer science. She earned her doctorate degree in Medical Sciences in 2008. As a postdoctoral researcher she worked within the Nordic Centre of Excellence in Disease Genetics and, later, The Cancer Risk Prediction Center. Since 2013 she is active at the Institute of Medicine, University of Gothenburg and pursuing her research interest in addressing nutritional imbalances in order to prevent disease.
Title: From population-based multigeneration data to nutritional research
Abstract: Population-based registers are a gold mine for studying familial aggregation of disease. However, incomplete data and missing family links pose methodological problems. Through my PhD work I focused on addressing such biases. My current research links nutritional deficiencies, obesity and chronic disease, as well as exploring the utility of old biobanks.
Samuli Ripatti, University of Helsinki
Samuli Ripatti, PhD, Professor in Biometry at the University of Helsinki and honorary faculty member of the Wellcome Trust Sanger Institute, is a statistical geneticist with a special interest in genetics of cardiovascular traits, lipids and metabolomics. Ripatti has been actively involved in large EU projects, including ENGAGE, BioSHaRE and CENTER-TBI, key player in multiple global complex disease genetics consortia, co-PI/steering group member in the key Finnish sequencing efforts, including SISu, Fin-Met-Seq and EUFAM, and the PI of the GeneRISK Study. He is an Associate Editor in PLoS Genetics. He is also chairing the Doctoral Programme in Population Health. He has published over 210 publications cited more than 13,000 times including first/last author papers in leading journals such as Nature Genetics, The Lancet, Biometrics and PLoS Genetics.
Title: Large-scale studies in complex disease genetics
Abstract: The past 10 years have revolutionized our understanding of the genetic factors related to the risks of complex diseases and to modification of their risk factors. Key advances behind this success include unbiased and relatively cheap genotyping and sequencing technologies, large-scale population-based study designs, proper control for confounding and false positive risks, reliable and efficient methods and implementations for variant calling, genotype imputation and association testing. Much of the advancements have been driven by the statistical genetics community, developing novel analysis strategies, methods and software. As more and more samples are now scanned using whole genome and whole exome sequencing technologies, we are now for the first time cataloguing near complete genetic variation in populations. This allows us to study the downstream health effects of low frequency and rare genetic variation and learn about the genetic architectures behind many complex diseases and traits. I will illustrate some of these new opportunities, designs and analysis methods using whole genome and whole exome sequencing in dyslipidemic families and in tens of thousands of population samples, and phenome-wide association studies linking genome-wide variant data to nationwide health registries.
Arvid Sjölander, Karolinska Institutet
Arvid Sjölander is Associate Professor of Biostatistics at the Department of Medical Epidemiology and Biostatistics at Karolinska Institutet. He mainly work with statistical methods in causal inference, and in a range of related areas such as doubly robust estimation, attributable fractions and sibling-comparison designs. He is also heavily involved in teaching, and in a number of applied epidemiological projects.
Title: Causal inference in theory and practice
Abstract: Causal inference is often considered to be a theoretical and difficult branch of statistics. In this presentation I will give examples from my own research, where I hope/believe that methods and thinking from causal inference have had an important practical influence.