- statslausanne - lists.sib.swiss

XXXIst ROES conference, September 9-12, 2019, Lausanne, Switzerland
by Rousson Valentin 25 Mar '19

25 Mar '19

Dear list members, This is to announce that the 31st conference of the Austro-Swiss Region (ROeS) of the International Biometric Society will take place in Lausanne, Switzerland, between September 9-12, 2019, under the motto: >From Controlled Trials to Big Data and Back Registration and abstract submission is open. Please visit the following link: http://wp.unil.ch/ibs-roes2019/ Statistical data analyses are sometimes classified as being either exploratory or confirmatory, while the reality of statistical practice often lies in between. This middle territory is exemplified by "model selection" issues and Frank Harrell's famous words: "Using the data to guide the data analysis is almost as dangerous as not doing so". The most accomplished confirmatory statistical analyses are conducted in the context of controlled (clinical) trials, where regulations and guidelines are to ensure a fully protocolled and planned statistical analysis. On the other hand, we are now living in the era of "big data" and "data science", where extreme forms of exploratory data analyses are encouraged with the hope that data quantity prevails over data quality. While data science is currently in vogue, there is also some perception that "those who ignore statistics are condemned to reinvent it", as Brad Efron once said. It might be a time to return from the big data paradigm towards more classical approaches and concerns, and to land somewhere between the two extremes of the purely confirmatory and purely exploratory data analyses. The XXXIst ROeS statistical conference will be a timely occasion to try to define what this "middle ground" should or could be to best meet the expectations of scientists. I hope to encounter many of you at this occasion! I would also be grateful if you could send this link and make publicity for the conference around your colleagues and within your network. With best wishes Valentin [1549450169707] Prof Valentin Rousson DFRI - Département Formation, Recherche et Innovation Route de la Corniche 10 | Bureau BIO02/02/103 CH-1010 Lausanne Tél direct : +41 (0)21 314 7328 www.iumsp.ch<http://www.iumsp.ch/>

1 0

Statistical forum on Thursday March 21 at 11:00
by Rousson Valentin 13 Mar '19

13 Mar '19

Dear all, Another seminar in our series of statistics talks for people interested in quantitative methods in biology and medicine will take place on Thursday March 21 2019 at 11:00 in the Salle Delachaux (SVA-Biopôle 2 building, first floor, Route de la Corniche 10, 1010 Lausanne, M2: Vennes). The speaker will be Dr. Marc Ryser, assistant professor at the Duke University School of Medicine, USA. Title : Estimating overdiagnosis in cancer screening studies: a statistical conversation about bias, identifiability and biologic mechanism Abstract : It is generally accepted that some screen-detected cancers are overdiagnosed and would not progress to symptomatic cancer if left untreated. However, because overdiagnosis is not directly observable, obtaining precise estimates constitutes a methodological challenge. In the first part of the talk we discuss the most common methods used to estimate overdiagnosis, namely excess-incidence and model-based approaches. We explain why most excess-incidence studies are prone to bias and consider model-based approaches that build on and extend methods for interval censored incidence as observed in cancer screening trials. Our extension permits preclinical tumors that are non-progressive but raises parameter identifiability issues. We discuss identifiability and model misspecification and propose concrete solutions to address these challenges. In the second part of the talk, we revisit the problem of cancer overdiagnosis through the lens of cancer biology. In particular, we seek to reconcile the current paradigm of cancer initiation and progression with the statistical overdiagnosis estimates from cancer screening studies and highlight the need for multi-scale approaches that integrate evidence from multiple sources. I am also joining two papers related to this talk. Hope to see you there! Valentin (for the organizers: Jean-Luc Bulliard, Zoltán Kutalik, Valentin Rousson, Frédéric Schütz) [1549450169707] Prof Valentin Rousson DFRI - Département Formation, Recherche et Innovation Route de la Corniche 10 | Bureau BIO02/02/103 CH-1010 Lausanne Tél direct : +41 (0)21 314 7328 www.iumsp.ch<http://www.iumsp.ch/>

1 0

Statistical forum tomorrow (Thursday November 29 at 11:00)
by Rousson Valentin 28 Nov '18

28 Nov '18

Dear all, Another seminar in our series of statistics talks for people interested in quantitative methods in biology and medicine will take place TOMORROW on Thursday 29 November 2018 at 11:00 in the Salle Delachaux (Biopôle 2, first floor, Route de la Corniche 10, 1010 Lausanne, M2: Vennes), which will be presented by Prof. Giuseppe Melfi from the University of Neuchâtel. For this last seminar of the year, note that the topic might be a bit unusual for us statisticians, since it will address the mathematical and fascinating concept of "weird numbers". Note also that the seminar will be held in French. Title: Résultats, problèmes et conjectures sur les nombres étranges Abstract: Un nombre n est dit étrange si la somme de ses diviseurs propres (y compris 1 mais pas n) est supérieure à n, et si n ne peut pas être exprimé comme une somme de certains de ses diviseurs propres. Par exemple, les huit plus petits nombres étranges sont 70<https://fr.wikipedia.org/wiki/70_(nombre)>, 836, 4030, 5830, 7192, 7912, 9272 et 10'430. En 1972, Benkoski et Erdös publient un article où ils démontrent certaines propriétés que nous exposeront dans ce séminaire, et proposent un certain nombre de conjectures et problèmes ouverts. L'un des problèmes les plus intéressants est de déterminer s'il existe des nombres étranges impairs. À l'heure actuelle tous les nombres étranges que l'on connait sont pairs, mais il n'y a pas de raisons évidentes qui empêcheraient l'existence de nombres étranges impairs. Nous allons voir que ces problèmes sont liés à l'étude des nombres étranges primitifs, c'est-à-dire des nombres étranges qui ne sont pas des multiples d'autres nombres étranges. Une approche qui semble prometteuse serait d'étudier les facteurs premiers de ces nombres étranges primitifs. Nous allons aussi aborder le problème de l'existence d'un nombre infini de nombres étranges primitifs, et le problème de déterminer combien de facteurs premiers un nombre étrange primitif peut contenir. Hope to see you there! Valentin (for the organizers: Zoltán Kutalik, Valentin Rousson, Frédéric Schütz) CHUV centre hospitalier universitaire vaudois Valentin ROUSSON - Professeur associé, mathematicien/statisticien Département universitaire de medecine et santé communautaires (DUMSC) Médecine sociale et préventive (IUMSP) Biostatistique et méthodes quantitatives BIO 2/2ème/103 Rte de la Corniche 10, CH - 1010 Lausanne +41 (0)21 314 73 28 TEL Valentin.Rousson(a)chuv.ch<mailto:Valentin.Rousson@chuv.ch> www.chuv.ch<http://www.chuv.ch/>

1 0

Statistical forum on Thursday November 8 at 11:00
by Rousson Valentin 23 Oct '18

23 Oct '18

Dear all, Another seminar in our series of statistics talks for people interested in quantitative methods in biology and medicine will take place on Thursday 8 November 2018 at 11:00 in the room Murray from the Aquatis Hôtel, route de Berne 148, 1010 Lausanne (M2 stop : Vennes). The speaker will be Prof. Elena Kulinskaya (School of computing Sciences, University of East Anglia, Norwich, UK). Title: Meta-analysis in practice: back to basics Abstract: Meta-analyses (MAs) generally follow a standard approach, established many years ago and codified in many guidelines and numerous software packages. The approach for random-effects MA involves the following steps: (a) Estimated variances of the study-level effects are substituted for the unknown true variances; (b) the between-study (heterogeneity) variance, tau-squared, is estimated by a generic method; (c) the estimate of the overall effect is obtained as a weighted average of the study-level effects, using inverse-variance weights; (d) the variance of the estimated overall effect is estimated by the reciprocal of the sum of the weights; and (e) a normal-theory confidence interval for the overall effect is centered at its estimate. Most of these steps involve assumptions that are usually not valid. Because of below-nominal coverage at step (e), the latest methodology publications recommend a confidence interval based on the t-distribution and a different estimate of the variance, developed by Hartung & Knapp (2001) and Sidik & Jonkman (2002). In this talk, I aim to move beyond two fallacies: (a) all effect measures can be analysed generically, and (b) estimated inverse-variance weights are the best way to combine study effects. I show that estimation of tau-squared should use methods specific to the effect measure. For estimating the overall effect, we advocate weights that do not involve estimated variances. We use fixed weights; i.e., the weights depend on nothing more than the studies' sample sizes, as suggested by Hedges and Olkin (1985), Hunter and Schmidt (1990), and many others. Specifically, we use the weights proportional to the effective sample size. This approach yields unbiased estimates of the overall effect with little loss of efficiency. To estimate tau-squared, we use improved approximations to the distribution of Cochran's Q for the particular effect measure. Such approximations were developed for the standardised mean difference (SMD) in Kulinskaya, Dollinger and Bjørkestøl (2011) and for the log-odds-ratio in Kulinskaya and Dollinger (2015). To achieve coverage close to the nominal 95%, we center the confidence interval for the overall effect at the fixed-weight estimate and use critical values from the t-distribution, in the spirit of Hartung & Knapp, and Sidik & Jonkman. In this talk, I present results of the extensive simulations comparing our approach with various methods that follow the standard MA approach focusing on the standardised mean difference and on the log-odds-ratio. This is a joint work with Ilyas Bakbergenuly and David Hoaglin Hope to see you there! Valentin (for the organizers: Zoltán Kutalik, Valentin Rousson, Frédéric Schütz) CHUV centre hospitalier universitaire vaudois Valentin ROUSSON - Professeur associé, mathematicien/statisticien Département universitaire de medecine et santé communautaires (DUMSC) Médecine sociale et préventive (IUMSP) Biostatistique et méthodes quantitatives BIO 2/2ème/103 Rte de la Corniche 10, CH - 1010 Lausanne +41 (0)21 314 73 28 TEL Valentin.Rousson(a)chuv.ch<mailto:Valentin.Rousson@chuv.ch> www.chuv.ch<http://www.chuv.ch/>

1 0

Statistical forum on Thursday October 18 at 11:00
by Rousson Valentin 12 Sep '18

12 Sep '18

Dear all, Another seminar in our series of statistics talks for people interested in quantitative methods in biology and medicine will take place on Thursday 18 October 2018 at 11:00, in the Salle Delachaux (Biopôle 2, first floor, Route de la Corniche 10, 1010 Lausanne, M2: Vennes), which I will present. Title: A five-decision testing procedure to infer the value of a unidimensional parameter Abstract: A statistical test can be seen as a procedure to produce a decision based on observed data, where some decisions consist of rejecting a hypothesis (yielding a significant result) and some do not, and where one controls the probability to make a wrong rejection at some pre-specified significance level. Whereas traditional hypothesis testing involves only two possible decisions (to reject or not a null hypothesis), Kaiser's directional two-sided test as well as the more recently introduced testing procedure of Jones and Tukey, each equivalent to running two one-sided tests, involve three possible decisions to infer the value of a unidimensional parameter. The latter procedure assumes that a point null hypothesis is impossible (e.g., that two treatments cannot have exactly the same effect), allowing a gain of statistical power. There are, however, situations where a point hypothesis is indeed plausible, for example, when considering hypotheses derived from Einstein's theories. In this article, we introduce a five-decision rule testing procedure, which combines the advantages of the testing procedures of Kaiser (no assumption on a point hypothesis being impossible) and Jones and Tukey (higher power), allowing for a non-negligible (typically 20%) reduction of the sample size needed to reach a given statistical power to get a significant result, compared to the traditional approach, while achieving "the abolition one and for all of the controversy over whether a one-sided or a two-sided test is appropriate", as L. Freedman once said. This is a joint work with Aaron McDaid and Zoltan Kutalik (IUMSP) (see the enclosed publication) Hope to see you there! Valentin (for the organizers: Zoltán Kutalik, Valentin Rousson, Frédéric Schütz) CHUV centre hospitalier universitaire vaudois Valentin ROUSSON - Professeur associé, mathematicien/statisticien Département universitaire de medecine et santé communautaires (DUMSC) Médecine sociale et préventive (IUMSP) Biostatistique et méthodes quantitatives BIO 2/2ème/103 Rte de la Corniche 10, CH - 1010 Lausanne +41 (0)21 314 73 28 TEL Valentin.Rousson(a)chuv.ch<mailto:Valentin.Rousson@chuv.ch> www.chuv.ch<http://www.chuv.ch/>

1 0

PhD position at University of Berne
by Rousson Valentin 13 Jun '18

13 Jun '18

Dear list members, The Institute of Social and Preventive Medicine at University of Berne is looking for a PhD student in Biostatistics and Epidemiology (contract for 3 years, start between September 2018 and January 2019, deadline for application: July 15, 2018). More information on: http://www.ispm.unibe.ch/unibe/portal/fak_medizin/ber_vkhum/inst_smp/conten… Best wishes Valentin CHUV centre hospitalier universitaire vaudois Valentin ROUSSON - Professeur associé, mathematicien/statisticien Département universitaire de medecine et santé communautaires (DUMSC) Médecine sociale et préventive (IUMSP) Biostatistique et méthodes quantitatives BIO 2/2ème/103 Rte de la Corniche 10, CH - 1010 Lausanne +41 (0)21 314 73 28 TEL Valentin.Rousson(a)chuv.ch<mailto:Valentin.Rousson@chuv.ch> www.chuv.ch<http://www.chuv.ch/>

1 0

Seminar "VIsual Trumpery", Thursday 28 June 2018 in Lausanne
by Frédéric Schütz 11 Jun '18

11 Jun '18

Dear all, this announce is a bit different from what we usually post in the "statslausanne" mailing-list, but will likely be of interest to many of you. On Thursday 28 June, Alberto Cairo will give a seminar on "Visual Trumpery": how are misleading charts and graphs being disseminated, and how to assess the reliability, truthfulness and accuracy of graphics. The presentation will take place in the Auditoire of the Maison de la Communication (Avenue de Florimont 1) at 17:30. You can find more details in the flyer attached. Attendance is free, but registration is compulsory at https://visualtrumperylausanne.eventbrite.com Let me know if you have any question. Hope to see you there ! Frédéric

1 0

Statistical forum on Tuesday May 15 at 11:00
by Rousson Valentin 26 Apr '18

26 Apr '18

Dear all, Another seminar in our series of statistics talks for people interested in quantitative methods in biology and medicine will take place on Tuesday 15 May 2018 at 11:00, in the Salle Delachaux (Biopôle 2, first floor, Route de la Corniche 10, 1010 Lausanne, M2: Vennes). The speaker will be Aziz Chaouch from the Institute of Social and Preventive Medicine, University of Lausanne. Title: Meta-analysis of heterogeneous predictive models based on Monte Carlo simulations and the concept of a meta-model Abstract: Predictive models are being increasingly used in clinical practice e.g. to predict future drug plasma concentrations in subpopulations of patients as a function of the dosage regimen. When different studies developed predictive models for the same outcome, one would like to combine evidence from these models to optimize the quality of predictions. One problem is that the different studies may have used different predictive models with e.g. different sets of predictors. We shall present a simulation approach to allow a meta-analysis of such heterogeneous predictive models. The method postulates a hypothetic model (called the meta-model) which supposedly approximates the data generating mechanism common to all studies. By generating pseudo-observations from the meta-model and refitting the possibly misspecified models to these pseudo-data, it is possible to estimate meta-model parameters using an iterative procedure. Statistical properties of the method will be presented in the context of linear and linear mixed models. This is a joint work with Thierry Buclin and Valentin Rousson Hope to see you there! Valentin (for the organizers: Zoltán Kutalik, Valentin Rousson, Frédéric Schütz) CHUV centre hospitalier universitaire vaudois Valentin ROUSSON - Professeur associé, mathematicien/statisticien Département universitaire de medecine et santé communautaires (DUMSC) Médecine sociale et préventive (IUMSP) Biostatistique et méthodes quantitatives BIO 2/2ème/103 Rte de la Corniche 10, CH - 1010 Lausanne +41 (0)21 314 73 28 TEL Valentin.Rousson(a)chuv.ch<mailto:Valentin.Rousson@chuv.ch> www.chuv.ch<http://www.chuv.ch/>

1 0

Another statistical forum on Monday April 16 at 15:00
by Rousson Valentin 03 Apr '18

03 Apr '18

Dear all, Another seminar in our series of statistics talks for people interested in quantitative methods in biology and medicine will take place on Monday 16 April 2018 at 15:00, in the Salle Delachaux (Biopôle 2, first floor, Route de la Corniche 10, 1010 Lausanne, M2: Vennes). The speaker will be Prof. Tianxi Cai from the Department of Biostatistics at the Harvard School of Public Health. Title: Enabling Imprecise EHR Data for Precision Medicine Abstract: While clinical trials remain a critical source for studying disease risk, progression and treatment response, they have limitations including the generalizability of the study findings to the real world and the limited ability to test broader hypotheses. In recent years, due to the increasing adoption of electronic health records (EHR) and the linkage of EHR with specimen bio-repositories, large integrated EHR datasets now exist as a new source for translational research. These datasets open new opportunities for deriving real-word, data-driven prediction models of disease risk and progression as well as unbiased investigation of shared genetic etiology of multiple phenotypes. Yet, they also bring methodological challenges. For example, obtaining validated phenotype information, such as presence of a disease condition and treatment response, is a major bottleneck in EHR research, as it requires laborious medical record review. A valuable type of EHR data is narrative free-text data. Extracting accurate yet concise information from the narrative data via natural language processing is also challenging. In this talk, I'll discuss various statistical and informatics methods that illustrate both opportunities and challenges when using EHR data for translational research. These methods will be illustrated using EHR data from Partner's Healthcare. Hope to see you there! Valentin (for the organizers: Zoltán Kutalik, Valentin Rousson, Frédéric Schütz) CHUV centre hospitalier universitaire vaudois Valentin ROUSSON - Professeur associé, mathematicien/statisticien Département universitaire de medecine et santé communautaires (DUMSC) Médecine sociale et préventive (IUMSP) Biostatistique et méthodes quantitatives BIO 2/2ème/103 Rte de la Corniche 10, CH - 1010 Lausanne +41 (0)21 314 73 28 TEL Valentin.Rousson(a)chuv.ch<mailto:Valentin.Rousson@chuv.ch> www.chuv.ch<http://www.chuv.ch/>

1 0

Statistical forum on Thursday April 12 at 11:00
by Rousson Valentin 03 Apr '18

03 Apr '18

Dear all, The next seminar in our series of statistics talks for people interested in quantitative methods in biology and medicine will take place on Thursday 12 April 2018 at 11:00, in the Salle Delachaux (Biopôle 2, first floor, Route de la Corniche 10, 1010 Lausanne, M2: Vennes). The speaker will be Elisabeth Dahlqwist, PhD Student at the « Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm », currently visiting Zoltan Kutalik's group at the IUMSP. Title: Regression standardization and attributable fraction estimation with between-within frailty models for clustered survival data Abstract: In epidemiological studies the relationship between an exposure and the time to some outcome is usually presented as a hazard ratio. However, the hazard ratio is a relative measure that does not give any information about the impact of the exposure on the time to the outcome on a population level. In public health and policy making interest is usually in how the prevalence of some disease can be reduced by some intervention targeted at reducing an exposure that we expect to cause the disease. The attributable fraction (or attributable risk) and standardization are useful measures in public health since it inform about the population impact of an exposure on an outcome. In this study we analyze the relationship between preterm birth (born before week 37 of gestation) and the time-to ADHD diagnosis. We expect that the relationship between preterm birth and ADHD is confounded by unobserved familial factors such as genes and childhood environment. In the presence of unobserved confounding we cannot estimate the causal effect of preterm birth on ADHD. A solution is to use family data and compare siblings born by the same parents but with discordant status in preterm birth. The main method that is used to adjust for family shared confounders in survival data is the stratified Cox model. However, the stratified Cox model cannot estimate absolute effects and thus, we cannot use the model to estimate standardized survival curves or the attributable fraction function. An alternative method is the frailty model (which parallels the random effect model for point outcomes). The limitation with the ordinary frailty model is that it does not adjust for family shared unobserved confounding. However, the between-within model adjust for family shared unobserved confounding and can be used for standardization. In this work we present how the between-within model can be used for standardization and the estimation of the attributable fraction. As an illustration of the difference between the stratified Cox model, the frailty model and the BW model we use data from Swedish registries in a within-mother-between-pregnancy analysis of preterm birth on the time-to-ADHD diagnosis/medication. We also estimate standardized survival curves and the attributable fraction function for preterm birth and ADHD. Hope to see you there! Valentin (for the organizers: Zoltán Kutalik, Valentin Rousson, Frédéric Schütz) CHUV centre hospitalier universitaire vaudois Valentin ROUSSON - Professeur associé, mathematicien/statisticien Département universitaire de medecine et santé communautaires (DUMSC) Médecine sociale et préventive (IUMSP) Biostatistique et méthodes quantitatives BIO 2/2ème/103 Rte de la Corniche 10, CH - 1010 Lausanne +41 (0)21 314 73 28 TEL Valentin.Rousson(a)chuv.ch<mailto:Valentin.Rousson@chuv.ch> www.chuv.ch<http://www.chuv.ch/>

1 0