Menu

Archives Séminaires Equipe Biostatistique

Archives de l'année en cours - 2017

  • Mardi 19 septembre 2017, à 13h00 - Amphi Louis - ISPED 

    Soufiane présentera : "Prediction models in high dimensional settings : a comparative study "

  • Mardi 05 septembre 2017, à 13h00 en salle Mann - ISPED

    Corentin présentera : "Score test for a random changepoint on a mixed model".


  • Jeudi 08 juin 2017 à 13h00, Amphi Louis - ISPED

    Il y aura 2 présentations d'étudiants de master actuellement en stage en Biostatistique à l'ISPED.

    Priscilla Andriamihaingo
    présentera : "Calibration of Recent Infection tests for estimating Human Immunodeficiency Virus (HIV) incidence in the African context : estimation of the Mean Duration of Recent Infection (MDRI) and the False Recent Rate (FRR)"

    et Bénédicte Driollet présentera : "Association entre la précarité et l’échec de transplantation rénale chez les enfants"


  • Mardi 06 juin 2017 à 14h00, Salle ED30 - ISPED

    Linda Valeri (Harvard) présentera :
    "Explaining the Total Effect in the Presence of Multiple Mediators and Interactions"

    Abstract :
    Mediation analysis allows decomposing a total effect into a direct effect of the exposure on the outcome and an indirect effect operating through a number of possible hypothesized pathways. A recent study has provided formal definitions of direct and indirect effects when multiple mediators are of interested. Parametric and semi-parametric methods to estimate path-specific effects have also been described. Investigating direct and indirect effects with multiple mediators can be challenging in the presence of multiple exposure-mediator and mediator-mediator interactions. Our study provides three main contributions: 1) we obtain counterfactual definitions of interaction terms when more than one mediator is present; 2) we derive a decomposition of the total effect that unifies mediation and interaction when multiple mediators are present; and 3) we illustrate the connection between our decomposition and the 4-way decomposition of the total effect introduced in the context of a single mediator. We employ the decomposition to investigate the interplay of adverse events and psychiatric symptoms in explaining the effect of antipsychotics on social functioning in schizophrenia patients.


  • Jeudi 1er juin 2017 à 14h00, Amphi Louis - ISPED

    Il y aura 2 présentations d'étudiants de master actuellement en stage en Biostatistique à l'ISPED :

    Florian Guillet présentera :
    "Evolution des marqueurs de progression de l'Atrophie Multi-Systématisée : analyse des sous-dimensions de l'échelle UMSARS"

    Benjamin Leblanc présentera :
    "Influence pronostique des progressions de cancer et évolution de la taille tumorale à partir d'essais cliniques en cancer". 


  • Mercredi 24 mai 2017 à 14h00, Amphi Louis - ISPED

    Il y aura 2 présentations d'étudiants de master actuellement en stage en Biostatistique à l'ISPED. 

    Marion Médeville présentera :
    "Evolution des performances cognitives dans la population âgée survivante : estimation suivant diverses hypothèses concernant la relation entre la mortalité et la réponse aux questionnaires"  

    Paul Tauzia présentera : 
    "Détection d'expression génique différentielle entre groupes de patients par ajustement sur les proportions cellulaires"

    Résumé : "Différents types cellulaires peuvent exprimer des gènes à des niveaux variables. On cherche ici à prendre en compte ces différences pour identifier les gènes dont le niveau d'expression varie significativement entre certaines populations. On compare une méthode de permutation publiée en 2010 à une approche originale qui utilise le t-test modifie de Smyth (très utilisé dans les analyses génomiques). On évaluera chacune des ces 2 approches grâce à une étude de simulations dans l'optique de l'appliquer in fine à des données issues d'un essais clinique pour un vaccin thérapeutique conte le VIH."

  • Jeudi 18 mai 2017 à 13h00, Amphi Louis, ISPED
    Il y aura 2 présentations d'étudiants de master actuellement en stage en Biostatistique à l'ISPED :

    Juan Naredo présentera "Dichotomization of continuous diagnostic tests using an imperfect gold standard: evaluation of the bias introduced in the diagnostic evaluation  by latent class models".

    - Louis Capitaine présentera : "Arbres et forêts aléatoires pour données longitudinales en grande dimension".


  • Mercredi 10 mai 2017 à 16h, Amphi Louis
    Denis Agniel (Rand Corporation, Harvard Medical School) présentera :
    "Prediction of complex phenotypes using semiparametric canonical correlation analysis"

  • Jeudi 13 avril 2017 à 14h30, Salle Mann, ISPED
    Cécile Proust-Lima présentera :
    "Joint modelling of multiple latent processes and clinical endpoints : Example in Alzheimer's disease "

      Résumé :
      Alzheimer's disease, the most frequent dementia in the elderly, is characterized by multiple progressive impairments in the brain structure and in clinical functions such as cognitive functioning and functional disability. Until recently, these components were mostly studied independently while they are fundamentally inter-related in the degradation process towards dementia.  We propose a joint model to describe the dynamics of multiple correlated latent processes which represent various domains impaired in the Alzheimer'€™s disease. Each process is measured by one or several markers, possibly non Gaussian. Rather than considering the associated time to dementia as in standard joint models, we assume dementia diagnosis corresponds to the passing above a covariate-specific threshold of a pathological process modelled as a combination of the domain-specific latent processes.  This definition captures the clinical complexity of dementia diagnosis but also benefits from an inference via Maximum Likelihood which does not suffer from the usual complications due to numerical approximations of multivariate integrals.  The method is illustrated on large French population-based cohorts of cerebral aging which include repeated information on brain structure (hippocampus volume, cortical signature) and/or clinical manifestations (cognitive functioning, physical dependency and depressive symptoms) as well as clinical-based diagnoses of Alzheimer's disease.



  • Jeudi 06 avril 2017 à 14h30, Salle ED 31, ISPED
    Julien Piscione présentera :
    "Liens entre charge de travail (matchs), performance individuelle et survenue de blessures après de la population des joueurs du Top 14 de rugby".


  • Jeudi 30 mars 2017 à 14h15, Amphi Louis, ISPED
    Lisa Belin présentera :
    "Adaptation des designs de phase II en cancérologie à un critère de jugement censuré"

    Résumé :
    La phase II d’un essai clinique représente une étape importante de l’évaluation d’une thérapeutique. Il s’agit d’une étape de sélection ayant pour objectif d’identifier les traitements efficaces, qui seront évalués de manière comparative en phase III, et ceux qui, jugés inefficaces seront abandonnés. Le choix du critère de jugement et la règle de décision sont les éléments essentiels de cette étape de l’essai clinique. En cancérologie, le critère de jugement est principalement de nature binaire (réponse au traitement). Cependant, le développement des molécules cytostatiques a remis en cause le choix du critère de jugement dans l’évaluation de ces molécules en particulier en phase II. La réponse tumorale a été critiquée au profit de la survie sans progression (PFS). Cette dernière est un critère de jugement censuré qui nécessite des plans d’expérience adaptés. Case et Morgan en 2003 puis Huang en 2010 ont répondu à cette problématique par une comparaison ponctuelle du taux de survie à un instant donné cliniquement pertinent à l’aide du test de Lin et al. Cependant, Kwak et Jung en 2013 a transposé le test du one-sample log-rank, jusqu’alors utilisé en épidémiologie, dans le contexte des essais phase II permettant ainsi de prendre en compte l’ensemble de l’information de la courbe de survie. Ces plans d’expérience permettent de mettre en place des essais de phase II à deux étapes mono bras, comparant une PFS observée à une PFS théorique-dite historique.

    Le schéma de Kwak et Jung permet d’intégrer le maximum d’information disponible et ainsi d’inclure moins de patients pour tester les mêmes hypothèses. Ce plan d’expérience nécessite cependant un suivi de tous les patients de leurs inclusions jusqu’à la fin de l’essai. Le design de Case et Morgan propose quant à lui de suivre tous les sujets jusqu’au temps d’intérêt clinique et ne recueille pas les événements survenant après ce temps d’intérêt. Lorsque l’hypothèse d’un suivi de tous les patients de leurs inclusions à la fin de l’essai semble irréaliste, nous avons proposé une adaptation du design de Kwak et Jung à un suivi réduit. Elle est basée sur le test du one-sample log-rank comme le design de Kwak et Jung mais utilise le même suivi que le design de Case et Morgan.

    Notre proposition permet de réduire le nombre de sujets nécessaire de 23% par rapport au design de Case et Morgan. Elle respecte les risques de première et de seconde espèce fixés dans le protocole. En cela, elle est séduisante et apparait comme une alternative au design de Kwak et Jung original qui nécessite un suivi parfait de l’ensemble des patients. Cette étude de simulation permet de mettre en évidence l’intérêt d’utiliser test du one-sample log-rank dans le contexte des essais de phase II et ses performances en présence d’un suivi réduit.

    1. Case LD, Morgan TM. Design of Phase II cancer trials evaluating survival probabilities. BMC Med Res Methodol. 2003;3:6.

    2. Huang B, Talukder E, Thomas N. Optimal Two-Stage Phase II Designs with Long-Term Endpoints. Stat Biopharm Res. 2010;2(1):51–61.

    3. Lin DY, Shen L, Ying Z, Breslow NE. Group sequential designs for monitoring survival probabilities. Biometrics. 1996;52(3):1033–41.

    4. Kwak M, Jung SH. Phase II clinical trials with time-to-event endpoints: Optimal two-stage designs with one-sample log-rank test. Stat Med. 2013.


  • Jeudi 23 mars 2017 à 13h00, Salle Mann, ISPED
    Mélanie Prague présentera :
    "Integrated approaches to analysis of cluster randomized trials: recent developments in marginal methods"

    Abstract : Whilst the standard parallel two-arm cluster randomized trials (CRTs) has been used for decades, there has been a range of innovative developments and alternatives proposed in the 12 years since the publication of the Murray et al. review. To cover this ground, Turner, Prague, Murray et al. wrote a series of two review papers on design and analysis of CRTs in 2017. I will present the conclusion of this review. I will particularly focus on marginal models and discuss their properties compared to conditional models.  After briefly discussing alternative semi-parametric approaches including quadratic inference functions (QIF) and targeted maximum likelihood (tMLE). I will focus on recently developed semi-parametric methodology based on generalized estimating equation (GEE) that simultaneously accounts for both baseline covariate imbalance and missing outcome data in CRTs (Prague et al. 2016, Biometrics). This approach has been developed in the causal inference framework using augmented and inverse probability weighted methods. It is doubly robust and easily implementable using a published R package called CRTgeeDR.

  • Jeudi 16 mars 2017 à 13h00, Salle ED 29, ISPED
    Camille Sabathé présentera :
    "Modélisation de l'effet de variables explicatives sur la probabilité de devenir dément : approche par pseudo-valeur"

  • Jeudi 09 février 2017 à 13h00, Salle Pous, ISPED
    Hélène Jacquin-Gadda présentera :
    "Régression quantile pondérée pour l'analyse de données longitudinales incomplètes tronquées par le décès : méthode et application à l'estimation de normes cognitives du sujet âgé à partir des données de Paquid ".

  • Jeudi 26 janvier 2017 à 13h00, Salle Mann, ISPED
    Boris Hejblum présentera :
    "Couplage probabiliste de dossiers médicaux informatisés à partir de codes diagnostiques en l’absence d’identificateur direct".

  • Jeudi 12 janvier 2017 à 13h00, Salle Mann - ISPED
    Loïc Ferrer présentera :
    "Comparison of landmarking and joint modelling approaches for dynamic predictions in presence of competing risks ".

  • Mardi 08 novembre 2016, à 13h en salle POUS
  • Bachirou Tadde présentera "Modélisation dynamique de la causalité entre processus latents : application aux sphères anatomique, cognitive et fonctionnelle dans la maladie d’alzheimer".

  • Mardi 15 novembre 2016, de 12h30 à 13h30, amphi Louis - ISPED
  • Ina Jazic (PhD Student at the Harvard University, Boston) présentera "Analysis of semi-competing risks data from a nested case-control study".

  • Mercredi 14 septembre 2016 à 13h, amphi Louis - ISPED
  • Cristian Meza, Professeur à l'Université de Valparaiso (Chili), membre du Centre for Research and Modeling of Random Phenomena - Valparaiso (CIMFAV) fera un exposé sur :
    "La classification de données longitudinales au travers d'un modèle semi-paramétrique avec effets mixtes et des estimateurs de type Lasso".

  • Mercredi 22 juin 2016 à 15h30, en salle 5 - ISPED
  • Stacia DeSantis (University of Texas) présentera :
    "Feature-Specific Penalized Joint latent Class Analysis for Genomic Data"

    Joint latent class models, i.e., models where indicators of latent class are jointly modeled with time to event endpoints, are commonly applied in cancer genomics. One concern about the fitting of such latent class models is the issue of model identifiability. When a large number of observed or manifest variables are present, standard likelihood based approaches for parameter estimation do not converge to a global maximum without the application of strict constraints. This problem is further exacerbated in the context of time to event endpoints, as well as in the context of grouping of genomic data into broad features of interest (for example, genomic features representing functionality of genes, location on the chromosome, copy number, etc). We build a series of penalized joint latent class models in the setting of moderate to high dimensional mixed categorical and continuous data types. Stable parameter estimation is accomplished through L_1 and L_2 penalization of the latent class model likelihood, as well as via Bayesian methods for variable selection and parameter shrinkage. We present an overview of our frequentist and Bayesian methods, simulation studies to assess bias and variance stabilization in moderate to high-dimensional settings, and apply the methods to several genomic data sets to illustrate their utility.

  •   Vendredi 27 mai 2016 à 11h00, en salle Pous - ISPED
  • Sebastien Haneuse (Harvard) présentera :  "Health Hierarchical models for semi-competing risks data with application to quality of end-of-life care for pancreatic cancer".

    Résumé : Readmission following discharge from an initial hospitalization is a key marker of quality of health care in the United States. For the most part, readmission has been studied among patients with `acute' health conditions, such as pneumonia and heart failure, with analyses based on a logistic-Normal generalized linear mixed model. Naive application of this model to the study of readmission among patients with `advanced' health conditions such as pancreatic cancer, however, is problematic because it ignores death as a competing risk. A more appropriate analysis is to imbed such a study within the semi-competing risks framework. To our knowledge, however, no comprehensive statistical methods have been developed for cluster-correlated semi-competing risks data. To resolve this gap in the literature we propose a novel hierarchical modeling framework for the analysis of cluster-correlated semi-competing risks data that permits parametric or non-parametric specifications for a range of components giving analysts substantial flexibility as they consider their own analyses. Estimation and inference is performed within the Bayesian paradigm since it facilitates the straightforward characterization of (posterior) uncertainty for all model parameters, including hospital-specific random effects. An efficient computational scheme, based on the Metropolis-Hastings-Green algorithm, is developed and had been implemented in the SemiCompRisks R package. The proposed framework is motivated by and illustrated with an on-going study of the risk of readmission among Medicare beneficiaries diagnosed with pancreatic cancer. Using data on n=5,298 patients at J=112 hospitals in the six New England states between 2000-2009, key scientific questions we consider include the role of patient-level risk factors on the risk of readmission and the extent of variation in risk across hospitals not explained by differences in patient case-mix.


  • Lundi 23 mai 2016 à 14h00, en salle 5 - ISPED
  • Corine Baayen (CHU de Nantes) présentera : "Evaluating effect of a drug using multiple candidate models for the dose-response relationship."

    Résumé : During development of a drug, typically the choice of dose is based on a Phase II dose-finding trial, where selected doses are included with placebo. Two common statistical dose-finding methods to analyze such trials are separate comparisons of each dose to placebo (using a multiple comparison procedure) or a model-based strategy (where a dose–response model is fitted to all data). The first approach works best when patients are concentrated on few doses, but cannot conclude on doses not tested. Model-based methods allow for interpolation between doses, but the validity depends on the correctness of the assumed dose–response model. Bretz et al. (2005, Biometrics 61, 738–748) suggested a combined approach, which selects one or more suitable models from a set of candidate models using a multiple comparison procedure. The method initially requires a priori estimates of any non-linear parameters of the candidate models, such that there is still a degree of model misspecification possible and one can only evaluate one or a few special cases of a general model. We propose an alternative multiple testing procedure, which evaluates a candidate set of plausible dose–response models against each other to select one final model. The method does not require any a priori parameter estimates and controls the Type I error rate of selecting a too complex model.


  •  Mardi 17 mai 2016 à 11h00, à l'Amphi Louis - ISPED

  • Jay Holden
    (University of Cincinnati) présentera : A Complex Network Account of Shape Changes in Response Time Distributions.

  • Jeudi 21 avril 2016, à 13h00, Amphi Louis - ISPED

    Anaïs Rouanet
    présentera : "COMPARISON OF MARGINAL AND CONDITIONAL ANALYSES FOR LONGITUDINAL DATA TRUNCATED BY DEATH AND DROPOUT".

    Résumé :
    In ageing studies, follow-up is often discontinued by death and/or dropout. In this framework, the two mainly used methods are mixed models, estimated by Likelihood Maximisation, and marginal models, estimated by Generalized Estimating Equations.
    It is admitted in the literature that mixed models provide subject-specific effects of covariates, conditional on the individual random effects. In the presence of death and dropout, we show that this interpretation still holds in the population currently alive, contradicting the idea that these estimators are only interpretable among a population with no risk of death (Kurland, 2009), as the likelihood maximisation procedure is equivalent to imputing data after death and dropout.
    Besides, marginal models provide marginal effects, averaged over the population. In the presence of death and dropout, they quantify the effect among the population currently observed, which may be selected if the death process is MAR/MNAR. Using weighting methods such as the inverse probability to be observed, given the subject is alive, or the inverse probability to be observed, we can estimate the effect in the population currently alive or in the population with no risk of dropout nor death, respectively.
    Within different frameworks of missing data, we compared the consistency and efficiency of marginal models and mixed models estimators through simulations. We also compared their estimated trajectories to check the interpretation of the regression parameters of each method. Finally, these methods were applied on the French cohort Paquid which includes psychometric tests of 3777 subjects followed every 2/3 years during 25 years.

  • Lundi 21 mars à 13h,  en salle 5 - ISPED

    Perrine Soret présentera : "Lasso-type methods for high-dimensional regression with censored outcome due to a limit of detection".


    <yun-hee> </yun-hee>
  • Lundi 14 mars à 16h (attention horaire inhabituel) en salle ED 52

    Vincent Audigier,  post doctorant à l'hôpital Saint-louis , présentera :
    "Imputation multiple par analyse factorielle".


    <yun-hee> </yun-hee>
  • Lundi 07 mars 2016, à 13h00 en salle Mann - ISPED

    Yun-Hee
    présentera : "Joint nested frailty models for recurrent screening visits and correlated survival data arising from Lynch Syndrome families".

  • Lundi 01 février à 13h en salle Mann - ISPED

    Loïc Ferrer
    présentera "Prédiction dynamique en présence d'évènements compétitifs : comparaison de diverses approches".



  • Lundi 11 janvier à 12h30 à 13h30 en salle Mann - ISPED

    Julien Asselineau présentera "Évaluation des performances diagnostiques des tests de détection de l'infection à Campylobacter par des modèles à classes latentes".

  • Mercredi 14 septembre 2016 à 13h, amphi Louis - ISPED
  • Cristian Meza, Professeur à l'Université de Valparaiso (Chili), membre du Centre for Research and Modeling of Random Phenomena - Valparaiso (CIMFAV) fera un exposé sur :
    "La classification de données longitudinales au travers d'un modèle semi-paramétrique avec effets mixtes et des estimateurs de type Lasso".
  • Mercredi 14 septembre 2016 à 13h, amphi Louis - ISPED
  • Cristian Meza, Professeur à l'Université de Valparaiso (Chili), membre du Centre for Research and Modeling of Random Phenomena - Valparaiso (CIMFAV) fera un exposé sur :
    "La classification de données longitudinales au travers d'un modèle semi-paramétrique avec effets mixtes et des estimateurs de type Lasso".
  • Mercredi 14 septembre 2016 à 13h, amphi Louis - ISPED
  • Cristian Meza, Professeur à l'Université de Valparaiso (Chili), membre du Centre for Research and Modeling of Random Phenomena - Valparaiso (CIMFAV) fera un exposé sur :
    "La classification de données longitudinales au travers d'un modèle semi-paramétrique avec effets mixtes et des estimateurs de type Lasso".

  • Vendredi 11 décembre, 11h30-12h30, salle Mann de l'ISPED

    Thème : A Bayesian spatio-temporal approach in the context of species distribution modelling

    Intervenant :
    David Conesa (Associate Professor of Biostatistics, "Spatial and Temporal Statistics in Epidemiology and Environment" Research Group, Universitat de Valencia)

    A l’occasion de son séjour à l’ISPED, Université de Bordeaux dans le cadre du programme de mobilité enseignante Erasmus+

  • Abstract :

    In this talk, I will review one of the most important tools of Bayesian analysis, namely the Bayesian hierarchical models. This kind of models are becoming so popular when one has to deal with many real situations in which information is presented in layers or hierarchies. Bayesian hierarchical models have been used largely but now they have become kind of necessary in all those practical complex situations because of their ability to deal with problems in the omics fields, Epidemiology, where it can be natural to have tens or hundreds of thousands of variables to be analysed. After reviewing the models, we will show some of the numerical approaches that have been introduced to deal with them (simulation methods such as MCMC, or Laplace approximations, such as INLA), and finally two practical settings (species distribution and detecting the onset of epidemics) where these models have been applied.

  • Lundi 23 novembre 2015 à 13h en Salle Mann - ISPED

    Thème : A Multi-Marker Genetic Association Test Based on the Rasch Model Applied to Alzheimer's Disease

    Intervenant : Wenjia Wang


  • Lundi 09 novembre 2015 à 13h en Salle Mann - ISPED

    Thème : Let's shine ! (Ou comment faire des documents web intéractifs avec le package R shiny)

    Intervenant :
    Robin Genuer


  • Jeudi 29 octobre 2015 à 10h à l'Amphi Louis - ISPED

    Thème : Group and sparse group partial least square approaches applied in genomics context.

    MOTIVATION: The association between two blocks of 'omics' data brings challenging issues in computational biology due to their size and complexity. Here, we focus on a class of multivariate statistical methods called partial least square (PLS). Sparse version of PLS (sPLS) operates integration of two datasets while simultaneously selecting the contributing variables. However, these methods do not take into account the important structural or group effects due to the relationship between markers among biological pathways. Hence, considering the predefined groups of markers (e.g. genesets), this could improve the relevance and the efficacy of the PLS approach. RESULTS: We propose two PLS extensions called group PLS (gPLS) and sparse gPLS (sgPLS). Our algorithm enables to study the relationship between two different types of omics data (e.g. SNP and gene expression) or between an omics dataset and multivariate phenotypes (e.g. cytokine secretion). We demonstrate the good performance of gPLS and sgPLS compared with the sPLS in the context of grouped data. Then, these methods are compared through an HIV therapeutic vaccine trial. Our approaches provide parsimonious models to reveal the relationship between gene abundance and the immunological response to the vaccine. AVAILABILITY AND IMPLEMENTATION: The approach is implemented in a comprehensive [Formula: see text] package called sgPLS available on the CRAN.

    Intervenant :
    Benoit Liquet


  • Lundi 19 octobre 2015 à 13h en salle Mann

    Thème : Non parametric tests for detecting changes in the dependence between the components of multivariate data, with or without change in marginal distributions.

    Intervenant :
    Tom Rohmer

  • Lundi 12 octobre 2015, Salle 5, ISPED - 13h

    Titre : " What should we use to study cognitive decline and assess its risk factors: composite scores or multivariate approaches? Results of a simulation study on PAQUID and MAP data ".

    Intervenante :
    Cécile Proust-Lima



  • Vendredi 03 juillet 2015, Amphi Louis, ISPED - 14h30

    Titre : "Evaluation d'un critère de substitution à la survie globale en oncologie"

    Intervenant : 
    Derek DINARD

    Titre : "Exemple d'implémentation d'un programme fortran sous un package R : modèle conjoint à fragilités"

    Intervenant : 
    Florin BUGA


  •   Mercredi 1er juillet 2015, Salle 5, ISPED - 11h

    Titre : "Développement d'un modèle conjoint à classes latentes pour l'étude des trajectoires d'exposition à l'amiante et le risque de survenu du mésothéliome pleural."

    Intervenant : 
    Amin El GAREH

  • Lundi 29 juin 2015, Amphi Louis, ISPED - 13h

    Titre : "Facteurs pronostiques environnementaux et comportementaux des accidents de la vie courante : comparaison de méthodes d'apprentissage statistique adaptées aux données de l'observatoire MAVIE ".

    Intervenante : 
    Marina TRAVANCA


    Titre : "Outils statistiques d’aide à la décision pour l’organisation de l’entraînement chez des sportifs de haut niveau. "

    Intervenante : 
    Gaëlle LEFORT


  • Lundi 15 juin 2015, Salle ED32, site Carreire - 13h

    Titre : "Étude du vieillissement cognitif par régression quantile longitudinale."

    Intervenant : 
    Darlin MBA


    Titre : "Comparaison de méthodes d’analyse de données longitudinales lorsque le suivi est tronqué par le décès."

    Intervenante : 
    Sylla DIENABOU


  • Mardi 26 mai 2015, Salle Mann, site Carreire - 13h

    Titre : "Transplantation rénale préemptive : Rôle dans la survie du greffon et du patient en France"

    Intervenante : 
    Mathilde PREZELIN


    Titre : "Modélisation dynamique des sphères cognitives et anatomiques dans la maladie d'Alzheimer"

    Intervenant : 
    Bachirou TADDE


  • Jeudi 21 mai 2015, Salle 5, site Carreire - 14h

    Titre : COMPARISON AND ASSESSMENT OF MODELS FOR PARTICLE DIFFUSION IN BIOLOGICAL FLUIDS"

    ABSTRACT:
    Rapidly progressing particle tracking techniques have revealed that foreign particles in biological fluids exhibit rich and at times unexpected behavior, with important consequences for drug delivery. Yet, there remains a frustrating lack of coherence in the description of these particles' motion.  Largely this is due to a reliance on functional statistics (e.g., mean-squared displacement) to perform model selection and assess goodness-of-fit.  However, not only are these functional characteristics typically estimated with substantial variability, but they are shared by many stochastic processes -- each making fundamentally different predictions for important quantities of scientific interest.

    Here, we conduct a detailed Bayesian analysis of leading candidate models for subdiffusive particle trajectories in human pulmonary mucus.  Model selection is achieved by way of intrinsic Bayes factors, which avoid both noninformative priors and "using the data twice".  Goodness-of-fit is evaluated via several second-order criteria along with exact model residuals.  Our findings suggest that a simple model of fractional Brownian motion describes the data just as well as a first-principles physical model of viscoelastic subdiffusion.

    Intervenant : 
    Martin LYSY


  • Jeudi 18 mai 2015, Salle 27B, site Carreire - 11h

    Titre : "Bayesian Model-Based Gating of Flow Cytometry Data"

    Intervenant : 
    Chariff ALKHASSIM


    Titre : "Modèle mathématique de l'effet de l'interleukine 7 sur les lymphocytes CD4 et le réservoir viral"

    Intervenante : 
    Laura VILLAIN


  • Lundi 11 mai 2015, Amphi Louis, ISPED - 14h

    Titre : "Summarizing the association between disease onset and survival under cross-sectional sampling"

    Abstract :
    In this work, we study a framework for quantifying the association between disease onset and survival. Usual models for bivariate times generally quantify departures from independence between the two times. If these times have a constrained sum, such as in the case of age at disease onset and residual lifetime from onset, model parameters may fail to have an appealing interpretation. To provide a an interpretable description of the association between disease onset and survival, we construct and study a novel semiparametric model. Using the idea of targeted maximum likelihood estimation, we then consider estimation and inference for parameters based on this model using survival data obtained through a cross-sectional survey with follow-up. The data generated by this common epidemiological design are subject to systematic biases. We infer the association between dementia and the residual lifetime of Canadian elderly individuals using data from the Canadian Study of Health and Aging.

    Intervenant : 
    Marco CARONE


  • Lundi 13 avril 2015, Salle 5, ISPED - 14h30

    Titre : "Diagnosing Misspecification of the Random-Effects Distribution in Mixed Models"

    Abstract:
    It is traditionally assumed that the random effects in mixed models follow a multi-variate normal distribution, making likelihood-based inferences more feasible theoretically and computationally.
    However, this assumption does not necessarily hold in practice, which may lead to biased and unreliable results, especially for inferences regarding the random-effects themselves. We introduce a novel diagnostic test based on the so-called gradient function proposed by Verbeke and Molenberghs (2013) to assess the random-effects distribution. We establish asymptotic properties of our test and show that, under a correctly specified random-effects distribution, the proposed test statistic converges to a weighted sum of independent chi-squared random variables each with one degree of freedom. The weights, which are eigenvalues of a square matrix, can be easily calculated. We also develop a parametric bootstrap algorithm for small samples. Our strategy can be used to check the adequacy of any distribution for random effects in a wide class of mixed models, including linear mixed models, generalized linear mixed models, and non-linear mixed models, with univariate as well as multivariate random effects. Both asymptotic and bootstrap proposals are evaluated via simulations and a real data analysis of a randomized multicenter study on toenail dermatophyte onychomycosis.

    Intervenant : 
    Reza DRIKVANDI


  • Lundi 16 mars 2015, Salle Mann, ISPED - 13h

    Titre : "Une méthode pour simuler des données longitudinale et d'histoire d'évènement dans le cadre d'un modèle joint".

    We present methods to simulate longitudinal and multi-state data under a joint model framework, by notably refering to the paper of M. J. Crowther (Simulating biologically plausible complex survival data, Statistics in Medecine, 2013).
    We discuss the encountered problems, especially when the baseline cumulative intensity functions are non-invertible and have not closed form expressions, and sugger solutions that require in particular integration and root-finding techniques.

    Intervenant : 
    Loïc FERRER

Consulter le programme actuel