Marie-Luce Taupin: using statistics to explain the living world

After completing a scientific baccalaureate, Marie-Luce Taupin embarked on a University General Studies Degree (DEUG) in mathematics applied to social sciences at Université de Nanterre in 1989. Then, in 1992, she joined Université Paris-Sud (now Université Paris-Saclay) to obtain an undergraduate degree and a master's degree in mathematics. In 1994, she completed a DEA (Master of Advanced Studies, equivalent to a 2nd year master's degree today) in stochastic modelling. In 1998, she defended her thesis at the same university, focusing on models with errors in variables.

Resulting from her research at the National Institute for Agricultural Research (INRA, now INRAE), this thesis examined estimation methods for models in which explanatory variables are not directly measurable, but are distorted by noise. The aim was to develop robust statistical methods for analysing phenomena based on uncertain data. Marie-Luce Taupin developed deconvolution methods to address this problem, which significantly complicates statistical analysis. "This technique aims to reduce noise in certain explanatory variables to make estimates more reliable."
 

From deconvolution to survival analysis

While teaching at the Université Paris-Descartes Technical Institute (IUT) (now the Paris Rives de Seine Technical Institute of Université Paris Cité), Marie-Luce Taupin continued her research at the Orsay Mathematics Laboratory (LMO - Univ. Paris-Saclay/CNRS) until 2006, before joining the Applied Mathematics at Paris 5 (MAP5) Laboratory at Université Paris-Descartes in 2007. During this time, she expanded her research into survival analysis, an essential area of biostatistics, which deals with censored data, such as in cases where patients are "lost to follow-up".

Ignoring such data can introduce bias into results. "Inaccurate measurements can cause us to overlook an important risk factor or underestimate the effectiveness of a treatment." The lecturer combined survival analysis with deconvolution and models with errors. This approach was instrumental, for example, when working with data from Hiroshima, to analyse the survival of people exposed to radiation, taking into account the uncertainty in radiation exposure levels.
 

Exploring high-dimensional data

Marie-Luce Taupin obtained her Accreditation to Supervise Research (HDR) in 2008. In 2009, she was appointed professor at Université d’Évry where she focused on high-dimensional estimation, particularly when analysing patient survival where there are numerous data, as for gene expression levels and associated censored durations. "The challenge lies in processing hundreds of thousands of variables for just a few thousand individuals." To identify the few truly influential variables, she used Lasso-type penalties, a method based on the sparsity principle, according to which only a fraction of the variables are significant.
 

INRA secondment: sensitivity analysis and simplified models for the environment

Seconded to INRA from 2012 to 2015, Marie-Luce Taupin took on a new challenge to simplify complex physical models. For instance, "simulating the diffusion of a pollutant requires so many parameters to be taken into account - soil type, presence of forest, farms, etc. - that each calculation can take several days." The lecturer developed simplified, reliable meta-models based on penalised criteria, a method which is particularly effective for analysing the diffusion of nitrogen products in soils.
 

From sensitivity analysis to multiple interactions

Since 2015, Marie-Luce Taupin's research has centred on two major areas. The first involves the construction of robust confidence intervals for R², an indicator measuring the proportion of variance explained by the variables in a model. "Our method remains reliable even without strict assumptions about the data, which greatly broadens its applications."

The second line of research explores the selection and analysis of interactions between variables in quadratic regression. For example, in a model studying the risk of cardiovascular disease, the combined effect of hypertension and smoking may be significant, even if each factor taken in isolation has little influence. This hierarchy between individual variables and their interactions is central to Marie-Luce Taupin's research, with practical applications, notably in cellular metabolic networks. "At the intersection of mathematics and the life sciences, precise statistical modelling can transform our understanding of complex phenomena."

An inclusive vision of education

As Director of the Mathematics Department at Université d’Évry from 2013 to 2021, Marie-Luce Taupin adopted an ambitious educational strategy geared towards professional integration. "Our mission is not just to teach future PhD candidates, but also to prepare the majority of our students for the job market." Through the personal study and integration project (“Projet personnel d’études et d’insertion” - PPEI), she encouraged students to explore non-academic professions related to mathematics, such as biostatistics or trading, by analysing the missions, skills and education required.

At the same time, she led the 1st year master's degree in Mathematics in Interaction from 2018 to 2023, before taking charge of the 2nd year master's degree in Data science in health, insurance and finance in 2024, a programme which she helped to establish. Her teaching, covering 2nd year undergraduate through to 2nd year master's degrees, takes into account the specific challenges faced by students at Université d’Évry. "Many of them face social difficulties, especially young international students who may feel isolated. Teaching here also means supporting them beyond mathematics."

Commitments to equality and the environment

Alongside her duties as a professor and researcher, Marie-Luce Taupin is actively involved in a number of key bodies. As a member of the National Council of Universities (CNU) in applied mathematics, she contributes to decisions on the qualifications and promotions of lecturers. Locally, she serves on the faculty council, where she takes part in discussions on positions and teaching models, an important role in a context of budget constraints. In 2025, she also became a member of the steering committee of the Fondation de Mathématiques Jacques Hadamard.

The lecturer's commitment is reflected in her active participation in her laboratory's gender equality and sustainable development committees. "Regarding equality, we are working to raise awareness among colleagues about stereotypes in recruitment and to support the next generations."

In February 2025, she is organising awareness days for 250 secondary school students, led by a sociologist. On the environmental front, Marie-Luce Taupin is helping to assess the laboratory's impact and rethink the policy of scientific missions. "These issues need to be incorporated into our collective decision-making."