Methodological Cross-Cutting Axis

Axe transversal méthodologique

Our work aims to develop, adapt, and apply innovative methodologies to address the complex challenges of environmental and climate health research. By leveraging interdisciplinary approaches drawn from epidemiology, data science, climate modeling, and the social sciences, our overarching goals are to:

  • strengthen the rigor and robustness of studies conducted within the Cités team,
  • drive methodological innovation to address the limitations of conventional approaches when confronted with the complexity of socio-environmental systems, and
  • promote the transfer of methodological knowledge and capacity building within the scientific community and among non-traditional research stakeholders.

Five methodological domains structure this cross-cutting axis

Adapting methods to heterogeneous contexts: Developing and validating approaches that allow research findings to be transported across different geographic, temporal, or population contexts, while accounting for variability in socioeconomic, climatic, and environmental determinants. This includes approaches based on optimal transport or doubly robust methods with super learner.

Strengthening and advancing causal inference methods: Mobilizing and adapting cutting-edge causal epidemiology methods (g-computation, inverse probability weighting, doubly robust estimators, instrumental variable methods) to identify cause-and-effect relationships between environmental exposures and health outcomes, while controlling for confounding factors and accounting for complex interaction mechanisms.

Leveraging satellite imagery for exposomics: Integrating remote sensing data and machine learning techniques to precisely characterize environmental exposures (air pollution, wildfire smoke, urban heat islands, green spaces) at spatial and temporal resolutions that are relevant to public health.

Advancing climate modeling applied to health: Using regional climate models and projections to anticipate the future health impacts of climate change, conduct attribution studies, and evaluate the effectiveness of adaptation strategies, while incorporating uncertainties and variability across climate scenarios.

Innovating with Ecological Momentary Assessment (EMA): Deploying real-time data collection methods through mobile technologies to capture environmental exposures, behaviors, and health states under everyday living conditions, enabling a dynamic and individualized approach to environmental epidemiology.

 

Three core principles underpin our methodological approach

  • Promoting rigor and transparency by systematically documenting methodological choices, fostering the reproducibility of analyses, and embracing the principles of open science.
  • Fostering interdisciplinarity by building bridges across disciplines (epidemiology, statistics, environmental sciences, geography, computer science) to develop integrated approaches suited to the complexity of the research questions at hand.
  • Ensuring knowledge transfer and training by sharing methodological developments through publications, tutorials, training sessions, and collaborations, in order to strengthen the capacity of the scientific community and support evidence-based decision-making.

Selected publications

Members of the axis

Dorcas Ebedi Nding Dorcas Ebedi Nding PhD student Compound climate events, Environmental health, Child health, Sub-Saharan Africa, Climate epidemiology Show email address
Pierre Scala Pierre Scala PhD student Causal inference, Environmental epidemiology, Quasi-experimental method, Heat wave Show email address
Robin Botrel Statistician, PhD student LC-MS data, Alignment, Functional approach, Pollutants, Statistical modeling Show email address
Tarik Benmarhnia Tarik Benmarhnia Co-leader of the Cités team | Epidemiologist, Professor at UCSD and EHESP Climate change, urban health, causal inference methods, environmental justice, policy evaluation Show email address
Valérie Gares Valérie Garès Professor Chair in Statistics Survival data analysis, Multi-source data integration, Functional data, Statistical learning, Application in healthcare Show email address