Past projects

APPLIED RESEARCH CONTRACTS

> SAS Partnership (2012-2014)

▪ FINANCING: SAS
▪ GRANT HOLDER: C. Legrand
The SAS software is one of the most used statistical software in the world. Since several years, there exist a partenariat between SAS and Institut de Statistique, Biostatistique et Sciences Actuarielles (ISBA) through which courses of programming in SAS and data mining techniques are organized. These courses are open to all master students as well as to PhD students and to all researchers of the UCL. Within the context of this partenariat, SAS also support (financially and logistically) the organisation of short courses within ISBA.

> Statistical monitoring applied to research trials (2011-2013)

▪ FINANCING: Biowin (pôle de compétitivité region wallonne)
▪ GRANT HOLDER: C. Legrand
The objective is to develop and validate a software for central statistical monitoring of clinical trial data quality. This software will ne primarily based on a large number of statistical tests as well as on techniques from data mining and artificial intelligence. This tool will allow to detect abnormalities in data, resulting from errors, negligence, or fraud.

> External Statistician of the Independent Data Monitoring Committee for GSK-Biologicals Study “MAGRIT trial MAGE-A3 as Adjuvant Non-Small Cell LunG CanceR ImmunoTherapy” (2009-2013)

▪ FINANCING: GlaxoSmithKline Biologicals
▪ GRANT HOLDER: C. Legrand
GlaxoSmithKline Biologicals is currently conducting a large phase III clinical trial in lung cancer aiming to investigate the effect of an antigen-specific cancer immunotherapeutic as adjuvent treatment for patients with resectable non-small cell lung cancer. This clinical trial will enroll more than 2000 patients and to ensure the safety of these patients, this trial is regularly monitored by an independent committee of experts (Independent Data Monitoring Committee). The role of this IDMC is to review, at regular time interval, all the data available to ensure that further continuation of the trial is ethical and eventually to make recommendations with regards to the conduct of the trial. This IDMC is composed of 5 experts; 4 medical doctors and 1 statistician. Catherine Legrand, Institut de Statistique, Biostatistique et Sciences Actuarielles (ISBA), acts as independent statistician in this IDMC.

> SAS Partnership (2008-2012)

▪ FINANCING: SAS
▪ GRANT HOLDER: C. Legrand
The SAS software is one of the most used statistical software in the world. Since several years, there exist a partenariat between SAS and Institut de Statistique, Biostatistique et Sciences Actuarielles (ISBA) through which courses of programming in SAS and data mining techniques are organized. These courses are open to all master students as well as to PhD students and to all researchers of the UCL. Within the context of this partenariat, SAS also support (financially and logistically) the organisation of short courses within ISBA.

RESEARCH PROJECTS UNDER CONTRACTS AND COOPERATION PROJECTS

European Union

> Semiparametric Inference for Complex and Structural Models in Survival Analysis – COSMOS – advanced grant (2016 – 2021)

▪ FINANCING: European Research Council under the European Community's Seventh Framework Programme 2008-2013
▪ GRANT HOLDER: I. Van Keilegom
In survival analysis investigators are interested in modeling and analysing the time until an event happens.
It often happens that the available data are right censored, which means that only a lower bound of the time of interest is observed. This feature complicates substantially the statistical analysis of this kind of data.
The aim of this project is to solve a number of open problems related to time-to-event data, that would represent a major step forward in the area of survival analysis.
These open problems are related to cure models, dependent censoring, measurement errors and endogeneity.

> M- and Z-estimation in semiparametric statistics: applications in various fields (2008-2014)

▪ FINANCING: European Research Council under the European Community's Seventh Framework Programme 2008-2013
▪ GRANT HOLDER: I. Van Keilegom
The area of semiparametric statistics is, in comparison to the areas of fully parametric or nonparametric statistics, relatively unexplored and still in full development. Semiparametric models offer a valid alternative for purely parametric ones, that are known to be sensitive to incorrect model specification, and completely nonparametric models, which often suffer from lack of precision and power. A drawback of semiparametric models so far is, however, that the development of mathematical properties under these models is often a lot harder than under the other two types of models. The present project tries to solve this difficulty partially, by presenting and applying a general method to prove the asymptotic properties of estimators for a wide spectrum of semiparametric models.
The objectives of this project are twofold. On one hand a general theory will be applied by Chen, Linton and Van Keilegom (2003) for a class of semiparametric Z-estimation problems, to a number of novel research ideas, coming from a broad range of areas in statistics. On the other hand it will be shown that some estimation problems are not covered by this theory. A more general class of semiparametric estimators (M-estimators called) will be considered and a general theory for this class of estimators will be developed. This theory will open new horizons for a wide variety of problems in semiparametric statistics.

> Estimating monotone boundaries and frontiers (2011-2013)

▪ FINANCING: Support for training and career development of researchers (Marie Curie) – Intra-European Fellowships (IEF)
▪ GRANT HOLDER and RESEARCHER: A. Daouia (post-doc.)
▪ PROMOTOR / COORDINATOR: I. Van Keilegom
The estimation of monotone support boundaries is relatively unexplored and still in full development. This research project examines two partial frontier models which provide a valid alternative for purely stochastic ones, that are known to be sensitive to model misspecification, and for completely envelopment models, which often suffer from lack of robustness and precision. The development of mathematical properties under these two recent models is, however, often a lot harder than under the other ones. This project tries to solve this difficulty by attacking many unsolved issues.
 

Belgium – Federal support - BSP (Belgian Science Policy)

Developing crucial statistical methods for understanding major complex dynamic systems in natural, biomedical and social sciences - IAP Phase VII (2012-2017)

Efficient rank-based estimation in semiparametric copula models (2014-2016)

▪ FINANCING: Projet de recherche (PDR), FNRS
▪ GRANT HOLDER: G. Mazo
▪ PROMOTOR: J. Segers
The joint distribution function of a random vector can be decomposed in the marginal distribution functions and the copula.
The copula describes the dependence structure in a margin-free way. Applications of dependence modelling via copulas cover a wide variety of fields, from finance and actuarial sciences to biology and medical sciences.
In a semiparametric copula model, the copula is assumed to belong to a parametric family, while the margins are unspecified. Inference on the copula parameter is an important problem which is not yet completely understood. Given the group structure of the model, it is natural to use rank-based estimators. The most popular rank-based estimator is the pseudo-likelihood estimator. However, it is known that in general, this estimator is not semiparametrically efficient. To arrive at an efficient rank-based estimator, two avenues will be explored: first, a one-step update estimator, and second, a maximum rank-likelihood estimator. The update technique has already been applied in the special case of Gaussian copula models. However, for general copula models, the update step is challenging, as it depends on the efficient score function, whose components are given as the solutions of a coupled system of Sturm-Liouville differential equations. The maximum rank-likelihood estimator exploits the idea that the reduction to ranks should involve no loss of information. In the case of Gaussian copula models, this method has shown good performance in Monte Carlo simulations. Its large-sample properties, however, remain to be discovered.

> Statistical analysis of association and dependence in complex data - IAP Phase VI (2007-2011)

▪ FINANCING: Interuniversity Attraction Pole Programme, Belgian Science Policy, Brussels, Belgium
▪ PROMOTOR: I. Van Keilegom
▪ PARTNER INSTITUTIONS: Katholieke Universiteit Leuven (Belgium); Universiteit Hasselt (Belgium); Universiteit Gent (Belgium)
▪ EUROPEAN PARTNERS: Université Joseph Fourier, Grenoble, France; Erasmus Medical Center, The Netherlands; Universidad de Santiago de Compostela, Spain; London School of Hygiene and Tropical Medicine, United Kingdom.
One key aim of statistics is to analyze in an appropriate way the dependence and association present in a dataset. The data that are collected nowadays to analyze these dependence structures are often of a complex nature and also the research questions are of an ever increasing complexity. This requires the construction of new models, or the adaptation of existing models, which is a challenging task. The development of new methods and intensive interaction between experts will also be required to cope with these complex data. The global objective of the network is to develop new models and methodological tools to do inference and to analyze these complex data structures. 

Website: http://sites.uclouvain.be/IAP-Stat-Phase-V-VI/PhaseVI/index.html

Belgium – French-speaking community support

> Stochastic Modelling of Dependence: Systems under Stress (2012-2017)

> Semiparametric inference for survival and cure models (2011-2016)

▪ FINANCING: Action de Recherche Concertée (ARC), Communauté Française de Belgique
▪ PROMOTORS: A. El Ghouch, C. Heuchenne, P. Lambert, C. Legrand, I. Van Keilegom (main promotor)
When modeling time-to-event data, we typically assume that all subjects are at risk and will experience the event of interest if followed long enough. However, a typical feature of most medical applications is the possibility of "cure", in the sense that some of the subjects will actually not experience the event. Cure models are survival models allowing a cured proportion of individuals. Moreover, measuring times to a certain event in practice naturally induces the presence of right censoring, meaning that one only observes lower bounds for these quantities.  In this project, we study and extend popular semiparametric regression models when the response is possibly right-censored and is allowed to correspond to a non experienced event. Beyond medicine, this type of problems is encountered in a wide variety of fields of applications, like sociology, economy, insurance, ecology, applied sciences, etc.

> Econometric modelling of multivariate financial time series (2007-2012) 

▪ FINANCING: Action de Recherche Concertée (ARC), Communauté Française de Belgique
▪ PROMOTORS: L. Bauwens, C. Hafner, J. Segers, R. von Sachs (main promotor)
This interdisciplinary research project deals with modelling, estimation and prediction of the dynamics and the temporal dependence in the mean and the variance-covariance structure of multivariate time series data arising in economic and financial applications. Particular emphasis is put on questions such as dimension reduction (factor approach, modelling of co-movements), non-stationary behaviour over time, modelling of structural breaks (regime-switching), volatilities with and without jump behaviour, etc.
These questions are addressed by a number of econometricians and statisticians using and comparing a series of modern approaches in parametric, semi-parametric and non-parametric statistics. Applications to real data will help to access the quality of the proposed models and estimation procedures.

> Specification of the frailty density in a frailty model and application to the analysis of data from multicenter clinical trials (2011-2014)

▪ FINANCING: Fonds pour la formation à la recherche dans l’industrie et dans l’agriculture (FRIA), FNRS
▪ GRANT HOLDER : M. Munda
▪ PROMOTOR: C. Legrand
Over the past years, frailty models became probably the most popular model to analyse clustered survival data. However, very little work has been done on the choice of the frailty density. In this project, we would like to investigate further the type of dependence induced by various choices of frailty density, the impact of misspecification of this density, and aim to develop diagnostics and goodness-of-fit tests. This research has direct application to the analysis of data from multicenter survival clinical trials. Indeed, frailty models may become the standard way to take the clustering in such multicenter datasets into account.

Belgium – Walloon region support

> Algorythmic governementality (2013-2017)

▪ FINANCING: Projet de recherche (PDR), FNRS
▪ PROMOTORS: Thomas Berns (ULB), Dominique Deprins (UCL (IMMAQ/ISBA) & Université Saint-Louis - Bruxelles), Antoinette Rouvroy (FUNDP)
The numerical capture of “reality”, and the statistical processes of the numerisation of the data play a role increasingly more essential in the whole of the normative practices, due to the fact of the public actors or the private actors. This new “cybernetic” modality of "control” is the fruit of a convergence of different factors: the evolution of the statistical algorithms, the explosion of the amount of available numerical data on the Internet, the development of a cognitive capitalism of « the attention », the politics focusing on the leitmotivs such as security, effectiveness, immediacy, interaction, interconnection, fluidity and comfort. This project will analyze the evolution of the normative actions to the light of the evolution of the statistical practices. The approach, nourished in an interdisciplinary way (philosophy, statistics, right, social sciences) is primarily analytical: far from a denounciative posture and concerned about the materiality of the statistical, technical and normative processes , we are rather interested to the question of understanding of what changes currently on the ground of the statistical practices that impose their omnipresence, of what they are constitutive of the contemporary gouvernementality, of what they involve changes in the nature itself of normativity and in the manner by which the human beings are normalized; we intend to do this in particular by observing at the same time the distance that these new practices induce with the legal normativity and the possible consecutive displacements of the subjectivation. By the analysis of the passage of the statistical government based on the classical probabilistic statistics, towards the decisional statistics implied in the automatic algorithmic systems (autonomic data-processing, intelligent environments) characterizing what we call the algorithmic gouvernementality, the purpose of our investigation is to shed light on the shifts induced by these new uses of the statistics on their manner of making the world meaningful, the way of anticipating the future, and the way in which, by the numerisation, the individuals are perceived and see themselve like subjects.

> Max-stable models for dependent extremes of random vectors in high dimensions (2013-2017)

▪ FINANCING: Fonds pour la formation à la recherche dans l’industrie et dans l’agriculture (FRIA), FNRS
▪ GRANT HOLDER: A. Kiriljouk
▪ PROMOTORS: M. Denuit and J. Segers

> Modelling extreme values of multivariate time Series (2013-2017)

▪ FINANCING: Mandat aspirant FNRS
▪ GRANT HOLDER: M. Warchol
▪ PROMOTORS: J. Segers and R. von Sachs
The overall aim of the project is to develop statistical theory and practice for extreme values of multivariate, regularly varying time series.
For stationary time series, estimators of the tail process will be constructed. The general theory will be specialized to Markov chains and linear processes, two widely applicable classes of models.

> Identification de nouveaux biomarqueurs non invasifs de l’endométrie par l’analyse des microARNs et de l’utilisation de la métabolomique – METABIOSE (2014-2017)

▪ FINANCING: SPW, Programme de recherche d’intérêt général ‘WB Health’, Convention n° 1318051
▪ PROMOTOR: B. Govaerts (for UCL)
The goal of this medical project is to find new biomarkers for endometriosis disease on the basis of microARNs and metabolomics technologies. The final goal is to develop a non-invasive medical detection kit of the biomarkers discovered in this research. In the project, the role of ISBA consists in developing and implementing adapted signal processing and statistical techniques to identify biomarkers from the various o-mic data generated in the project (microARNs and 1D and 2D H-NMR metabolic spectra).

> Mécanismes assurantiels ou de mutualisation des risques agricoles en Région wallonne (2011)

▪ Financing: Marché de services D31-1251, Services de recherche et de développement, DGARNE, SPW
▪ Grant holders: B. Henry de Frahan (ELI-A, UCL), C. Saegerman (UREAR, ULg), M. Denuit & P. Devolder (ISBA, UCL), B. Dubuisson (PJPR, UCL)
This project aims to study the risk management tools for stabilizing the revenue in the agricultural sector.

University support

> Adaptive nonparametric Bayesian estimation in inverse problems (2010-2012 et 2012-2014)

▪ FINANCING: Fonds Spéciaux de Recherche (FSR)
▪ GRANT HOLDER: J. Johannes
The objective of the project is the development of adaptive nonparametric Bayesian models for ill-posed inverse problems with noise in the operator. More precisely, we intend to study lower bounds for a-posteriori concentration rates, to construct prior distributions allowing to attain those and to compare the results with the minimax theory for adaptive estimation in purely frequentist models for ill-posed inverse problems.

> Study of different cure models in regression: construction of semiparametric inferential methods adapted to the available information and to the complex relations assumed between variables (2011-2014)

▪ FINANCING: Fonds de la recherche fondamentale collective (FRFC), FNRS
▪ PROMOTORS: A. El Ghouch, C. Heuchenne, C. Legrand, I. Van Keilegom (main promotor)

> Adaptive nonparametric Bayesian estimation in inverse problems (2010-2012)
▪ FINANCING: Fonds Spéciaux de Recherche (FSR)
▪ GRANT HOLDER: J. Johannes
The objective of the project is the development of adaptive nonparametric Bayesian models for ill-posed inverse problems with noise in the operator. More precisely, we intend to study lower bounds for a-posteriori concentration rates, to construct prior distributions allowing to attain those and to compare the results with the minimax theory for adaptive estimation in purely frequentist models for ill-posed inverse problems.

Foreign

> Metodología y Aplicaciones en Estadística Semiparamétrica, Funcional y Espacio Temporal (2009-2013)

▪ FINANCING: Spanish Ministry of Education and Science
▪ ASSOCIATE PARTNERS: I. Van Keilegom (main partner: W. González-Manteiga, Universidad de Santiago de Compostela, Spain)

> Efficiency and productivity of an industry (2009-2011)
▪ GRANT HOLDER: L. Simar
▪ PARTNER INSTITUTION: GREMAQ, Toulouse School of Economics, Agence Nationale de la Recherche (ANR), France

Private

Health insurance and longevity (2014-2016)

▪ FINANCING: DKV Belgium Chair
▪ GRANT HOLDER: P. Devolder
Development of actuarial and financial techniques for the pricing, hedging and reserving of health, disability and long term care insurances ; analysis of the influence of the longevity risk on these products.

> Le financement des pensions (2011-2015)

▪ FINANCING: Generali Chair
▪ GRANT HOLDER: P. Devolder
The purpose of the research is to develop actuarial and financial models in order to estimate the impact of various scenarios of reform of the pension systems in Belgium.
In particular, stochastic models will be considered in order to establish actuarial balance sheets of the social security pension system.  The longevity risk in particular will be deeply analyzed and modeled.

> Pension valuation and solvency (2013-2016)

▪ FINANCING: AG Insurance Chair
▪ GRANT HOLDER: P. Devolder
Development of a coherent and universal model of valuation and solvency requirement of pension liabilities for pension funds and insurance companies in a stochastic environment.

> Risk management for energy markets (2010-2014)

▪ FINANCING: GDF Suez Chair
▪ GRANT HOLDER: P. Devolder
The project aims to develop research in risk management applied to energy markets.
The main purpose is to develop new tools for commodity pricing based on stochastic techniques in continuous time especially applied to gas and electricity markets. LEVY processes will be the central tool in this perspective.
Applications to valuation of electricity derivatives will also be considered.

> Identification of a statistical model for the estimation of the vaccine efficacy of an influenza vaccine (2010-2014)

▪ FINANCING: GSK biologicals Chair
▪ GRANT HOLDER: C. Legrand
The project aims to identify and to develop optimal statistical techniques for the design and the analysis of the efficacy clinical trials of anti-flu vaccines, in collaboration with GSK Biologicals. The identified best approach and the motivation of this selection will be published in statistical journals so that peer-reviewed papers including academic authors will be available to be used as reference in the statistical analysis plan to be submitted and discussed with regulatory agencies.