2025Activity report​​​‌Project-TeamPREMEDICAL

7.1.1 declearn

• Keyword:
  Federated​​ learning
• Scientific Description:

  declearn​​​‌ is a python package​ providing with a framework​‌ to perform federated learning,​​ i.e. to train machine​​​‌ learning models by distributing​ computations across a set​‌ of data owners that,​​ consequently, only have to​​​‌ share aggregated information (rather​ than individual data samples)​‌ with an orchestrating server​​ (and, by extension, with​​​‌ each other).

  The aim​ of declearn is to​‌ provide both real-world end-users​​ and algorithm researchers with​​​‌ a modular and extensible​ framework that:

  (1) builds​‌ on abstractions general enough​​ to write backbone algorithmic​​​‌ code agnostic to the​ actual computation framework, statistical​‌ model details or network​​ communications setup

  (2) designs​​​‌ modular and combinable objects,​ so that algorithmic features,​‌ and more generally any​​ specific implementation of a​​​‌ component (the model, network​ protocol, client or server​‌ optimizer...) may easily be​​ plugged into the main​​​‌ federated learning process -​ enabling users to experiment​‌ with configurations that intersect​​ unitary features

  (3) provides​​​‌ with functioning tools that​ may be used out-of-the-box​‌ to set up federated​​ learning tasks using some​​​‌ popular computation frameworks (scikit-​ learn, tensorflow, pytorch...) and​‌ federated learning algorithms (FedAvg,​​ Scaffold, FedYogi...)

  (4) provides​​​‌ with tools that enable​ extending the support of​‌ existing tools and APIs​​ to custom functions and​​ classes without having to​​​‌ hack into the source‌ code, merely adding new‌​‌ features (tensor libraries, model​​ classes, optimization plug-ins, orchestration​​​‌ algorithms, communication protocols...) to‌ the party.

  Parts of‌​‌ the declearn code (Optimizers,...)​​ are included in the​​​‌ FedBioMed software.

  At the‌ moment, declearn has been‌​‌ focused on so-called "centralized"​​ federated learning that implies​​​‌ a central server orchestrating‌ computations, but it might‌​‌ become more oriented towards​​ decentralized processes in the​​​‌ future, that remove the‌ use of a central‌​‌ agent.

• Functional Description:

  This​​ library provides the two​​​‌ main components to perform‌ federated learning:

  (1) the‌​‌ client, to be run​​ by each participant, performs​​​‌ the learning on local‌ data et releases only‌​‌ the result of the​​ computation

  (2) the server​​​‌ orchestrates the process and‌ aggregates the local models‌​‌ in a global model​​

• News of the Year:​​​‌
  Two major releases with‌ key new functionalities including‌​‌ algorithms for group fairness​​ and the ability to​​​‌ use secure aggregation.
• URL:‌
  https://­gitlab.­inria.­fr/­magnet/­declearn/­declearn2
• Contact:
  Aurélien Bellet‌​‌
• Participants:
  Paul Andrey, Aurélien​​ Bellet, Nathan Bigaud, Marc​​​‌ Tommasi, Nathalie Vauquier
• Partner:‌
  CHRU Lille

7.1.2 CaMeA‌​‌

• Name:
  Causal Meta-Analysis for​​ Aggregated Data
• Keywords:
  Causality,​​​‌ Randomised control trials
• Functional‌ Description:
  Based on results‌​‌ from multiple clinical trials​​ (contingency tables cross-tabulating treatment​​​‌ and response), CaMeA measures‌ the effect of a‌​‌ treatment or intervention using​​ various metrics, such as​​​‌ Risk Ratio, Risk Difference,‌ and others.
• Publication:
  hal-05086779‌​‌
• Contact:
  Julie Josse
• Participants:​​
  Julie Josse, Clement Berenfeld​​​‌

7.1.3 missMDA

• Keyword:
  Missing‌ data
• Functional Description:
  The‌​‌ missMDA package is dedicated​​ to missing values in​​​‌ and with Multivariate Data‌ Analysis. It allows one‌​‌ to apply PCA, MCA,​​ FAMD and MFA on​​​‌ incomplete data. It performs‌ single and multiple imputation‌​‌ for continuous, categorical and​​ mixed data based on​​​‌ principal components methods
• URL:‌
  http://­factominer.­free.­fr/­missMDA/­index.­html
• Contact:
  Julie Josse‌​‌
• Partner:
  AGROCAMPUS

7.1.4 factominer​​

• Keywords:
  Dimensionality reduction, PCA,​​​‌ Text mining, Clustering
• Functional‌ Description:

  The FactoMineR package‌​‌ is dedicated to performing​​ principal components methods to​​​‌ explore, sum­-up and visualize‌ data. Dimensionality reduction methods‌​‌ include PCA, correspondence analysis​​ (CA) for count data​​​‌ such as documents-­words data,‌ multiple correspondence analysis (MCA)‌​‌ for categorical data such​​ as survey data, factorial​​​‌ analysis of mixed data‌ (FAMD) for both types‌​‌ of variables as well​​ as methods for groups​​​‌ of variables, of individuals‌ (multiple factorial analysis, MFA),‌​‌ for hierarchy …

  References:​​ https://husson.github.io/MOOC_AnaDo/index.html https://husson.github.io/MOOC.html#PCAcourse

• URL:
  http://­factominer.­free.­fr/­index_fr.­html​​​‌
• Contact:
  Julie Josse
• Partner:‌
  AGROCAMPUS

Results: Causal Meta​ Analysis: Rethinking the Foundations​‌ of Evidence-Based Medicine 59​​

Participants: Julie Josse,​​​‌ Clement Berenfeld, Ahmed​ Boughdiri, Remi Khellaf​‌, Aurélien Bellet.​​

Meta-analysis, by synthesizing effect​​​‌ estimates from multiple studies​ conducted in diverse settings,​‌ stands at the top​​ of the evidence hierarchy​​​‌ in clinical research. Yet,​ conventional approaches based on​‌ fixed- or random-effects models​​ lack a causal framework,​​​‌ which may limit their​ interpretability and utility for​‌ public policy. Incorporating causal​​ inference reframes meta-analysis as​​​‌ the estimation of well-defined​ causal effects on clearly​‌ specified populations, enabling a​​ principled approach to handling​​​‌ study heterogeneity. We show​ that classical meta-analysis estimators​‌ have a clear causal​​ interpretation when effects are​​​‌ measured as risk differences.​ However, this breaks down​‌ for nonlinear measures like​​ the risk ratio and​​​‌ odds ratio. To address​ this, we introduce novel​‌ causal aggregation formulas that​​ remain compatible with standard​​​‌ meta-analysis practices and do​ not require access to​‌ individual-level data. To evaluate​​ real-world impact, we apply​​​‌ both classical and causal​ meta-analysis methods to 500​‌ published meta-analyses. While the​​ conclusions often align, notable​​​‌ discrepancies emerge, revealing cases​ where conventional methods may​‌ suggest a treatment is​​ beneficial when, under a​​​‌ causal lens, it is​ in fact harmful.

Results:​‌ A Unified Framework for​​ the Transportability of Population-Level​​​‌ Causal Measures 62

Participants:​ Julie Josse, Clement​‌ Berenfeld, Ahmed Boughdiri​​.

Generalization methods offer​​​‌ a powerful solution to​ one of the key​‌ drawbacks of randomized controlled​​ trials (RCTs): their limited​​​‌ representativeness. By enabling the​ transport of treatment effect​‌ estimates to target populations​​ subject to distributional shifts,​​​‌ these methods are increasingly​ recognized as the future​‌ of meta-analysis, the current​​ gold standard in evidence-based​​​‌ medicine. Yet most existing​ approaches focus on the​‌ risk difference, overlooking the​​ diverse range of causal​​​‌ measures routinely reported in​ clinical research. Reporting multiple​‌ effect measures-both absolute (e.g.,​​ risk difference, number needed​​​‌ to treat) and relative​ (e.g., risk ratio, odds​‌ ratio)-is essential to ensure​​ clinical relevance, policy utility,​​​‌ and interpretability across contexts.​ To address this gap,​‌ we propose a unified​​ framework for transporting a​​​‌ broad class of first-moment​ population causal effect measures​‌ under covariate shift. We​​ provide identification results under​​​‌ two conditional exchangeability assumptions,​ derive both classical and​‌ semiparametric estimators, and evaluate​​ their performance through theoretical​​​‌ analysis, simulations, and real-world​ applications. Our analysis shows​‌ the specificity of different​​ causal measures and thus​​​‌ the interest of studying​ them all: for instance,​‌ two common approaches (one-step,​​ estimating equation) lead to​​​‌ similar estimators for the​ risk difference but to​‌ two distinct estimators for​​ the odds ratio.

Results:​​​‌ Rethinking the Win Ratio:​ A Causal Framework for​‌ Hierarchical Outcome Analysis. 63​​

Participants: Julie Josse,​​ Mathieu Even.

For​​​‌ hierarchical multivarariate outcomes, the‌ FDA recommends the Win‌​‌ Ratio and Generalized Pairwise​​ Comparisons approaches 84,​​​‌ 75. However, as‌ far as we know,‌​‌ these empirical methods lack​​ causal or statistical foundations​​​‌ to justify their broader‌ use in recent studies.‌​‌ To address this gap,​​ we establish causal foundations​​​‌ for hierarchical comparison methods.‌ We define related causal‌​‌ effect measures, and highlight​​ that depending on the​​​‌ methodology used to compute‌ Win Ratio, the causal‌​‌ estimand targeted can be​​ different, as proved by​​​‌ our consistency results, which‌ may then lead to‌​‌ reversed and incorrect treatment​​ recommendations in heterogeneous populations,​​​‌ as we illustrate through‌ striking examples. In order‌​‌ to compensate for this​​ fallacy, we introduce a​​​‌ novel, individual-level yet identifiable‌ causal effect measure that‌​‌ better approximates the ideal,​​ non-identifiable individual-level estimand. We​​​‌ prove that computing Win‌ Ratio or Net Benefits‌​‌ using a Nearest Neighbor​​ pairing approach between treated​​​‌ and controlled patients, an‌ approach that can be‌​‌ seen as an extreme​​ form of stratification, leads​​​‌ to estimating this new‌ causal estimand measure. We‌​‌ extend our methods to​​ observational settings via propensity​​​‌ weighting, distributional regression to‌ address the curse of‌​‌ dimensionality, and a doubly​​ robust framework. We prove​​​‌ the consistency of our‌ methods, and the double‌​‌ robustness of our augmented​​ estimator. These methods are​​​‌ straightforward to implement, making‌ them accessible to practitioners.‌​‌ Finally, we validate our​​ approach using synthetic data​​​‌ and on CRASH-3 [CRASH‌ et al., 2019], a‌​‌ major clinical trial focused​​ on assessing the effects​​​‌ of tranexamic acid in‌ patients with traumatic brain‌​‌ injury.

Results: Federated Causal Inference​​​‌ from Multi-Site Observational Data‌ via Propensity Score Aggregation.‌​‌ 66

Participants: Remi Khellaf​​, Aurélien Bellet,​​​‌ Julie Josse.

Causal‌ inference typically assumes centralized‌​‌ access to individual-level data.​​ Yet, in practice, data​​​‌ are often decentralized across‌ multiple sites, making centralization‌​‌ infeasible due to privacy,​​ logistical, or legal constraints.​​​‌ We address this problem‌ by estimating the Average‌​‌ Treatment Effect (ATE) from​​ decentralized observational data via​​​‌ a Federated Learning (FL)‌ approach, allowing inference through‌​‌ the exchange of aggregate​​ statistics rather than individual-level​​​‌ data. We propose a‌ novel method to estimate‌​‌ propensity scores by computing​​ a federated weighted average​​​‌ of local scores with‌ Membership Weights (MW)—probabilities of‌​‌ site membership conditional on​​ covariates—which can be flexibly​​​‌ estimated using parametric or‌ non-parametric classification models. Unlike‌​‌ density ratio weights (DW)​​ from the transportability and​​​‌ generalization literature, which either‌ rely on strong modeling‌​‌ assumptions or cannot be​​ implemented in FL, MW​​​‌ can be estimated using‌ standard FL algorithms and‌​‌ are more robust, as​​ they support flexible, non-parametric​​​‌ models—making them the preferred‌ choice in multi-site settings‌​‌ with strict data-sharing constraints.​​ The resulting propensity scores​​​‌ are used to construct‌ Federated Inverse Propensity Weighting‌​‌ (Fed-IPW) and Augmented IPW​​ (Fed-AIPW) estimators. Unlike meta-analysis​​​‌ methods, which fail when‌ any site violates positivity,‌​‌ our approach leverages heterogeneity​​ in treatment assignment across​​​‌ sites to improve overlap.‌ We show that Fed-IPW‌​‌ and Fed-AIPW perform well​​​‌ under site-level heterogeneity in​ sample sizes, treatment mechanisms,​‌ and covariate distributions. Both​​ theoretical analysis and experiments​​​‌ on simulated and real-world​ data highlight their advantages​‌ over meta-analysis and related​​ methods.

Results: Generalization under​​​‌ Byzantine & Poisoning Attacks:​ Tight Stability Bounds in​‌ Robust Distributed Learning 61​​

Participants: Thomas Boudou,​​​‌ Aurélien Bellet.

Robust​ distributed learning algorithms aim​‌ to maintain good performance​​ in distributed and federated​​​‌ settings, even in the​ presence of misbehaving workers.​‌ Two primary threat models​​ have been studied: Byzantine​​​‌ attacks, where misbehaving workers​ can send arbitrarily corrupted​‌ updates, and data poisoning​​ attacks, where misbehavior is​​​‌ limited to manipulation of​ local training data. While​‌ prior work has shown​​ comparable optimization error under​​​‌ both threat models, a​ fundamental question remains open:​‌ How do these threat​​ models impact generalization? Empirical​​​‌ evidence suggests a gap​ between the two threat​‌ models, yet it remains​​ unclear whether it is​​​‌ fundamental or merely an​ artifact of suboptimal attacks.​‌ In this work, we​​ present the first theoretical​​​‌ investigation into this problem,​ formally showing that Byzantine​‌ attacks are intrinsically more​​ harmful to generalization than​​​‌ data poisoning.

Results:​‌ Model Agnostic Differentially Private​​ Causal Inference. 67

Participants:​​​‌ Aurélien Bellet, Julie​ Josse, Christian Janos​‌ Lebeda, Mathieu Even​​.

Estimating causal effects​​​‌ from observational data is​ essential in fields such​‌ as medicine, economics and​​ social sciences, where privacy​​​‌ concerns are paramount. We​ propose a general, model-agnostic​‌ framework for differentially private​​ estimation of average treatment​​​‌ effects (ATE) that avoids​ strong structural assumptions on​‌ the data-generating process or​​ the models used to​​​‌ estimate propensity scores and​ conditional outcomes. In contrast​‌ to prior work, which​​ enforces differential privacy by​​​‌ directly privatizing these nuisance​ components and results in​‌ a privacy cost that​​ scales with model complexity,​​​‌ our approach decouples nuisance​ estimation from privacy protection.​‌ This separation allows the​​ use of flexible, state-of-the-art​​​‌ black-box models, while differential​ privacy is achieved by​‌ perturbing only predictions and​​ aggregation steps within a​​​‌ fold-splitting scheme with ensemble​ techniques. We instantiate the​‌ framework for three classical​​ estimators – the G-formula,​​​‌ inverse propensity weighting (IPW),​ and augmented IPW (AIPW)​‌ – and provide formal​​ utility and privacy guarantees.​​​‌ Empirical results show that​ our methods maintain competitive​‌ performance under realistic privacy​​ budgets. We further extend​​​‌ our framework to support​ meta-analysis of multiple private​‌ ATE estimates. Our results​​ bridge a critical gap​​​‌ between causal inference and​ privacy-preserving data analysis.

Results:​‌ Privacy Amplification Through Synthetic​​ Data: Insights from Linear​​​‌ Regression 53

Participants: Aurélien​ Bellet.

Synthetic data​‌ inherits the differential privacy​​ guarantees of the model​​​‌ used to generate it.​ Additionally, synthetic data may​‌ benefit from privacy amplification​​ when the generative model​​​‌ is kept hidden. While​ empirical studies suggest this​‌ phenomenon, a rigorous theoretical​​ understanding is still lacking.​​​‌ In this paper, we​ investigate this question through​‌ the well-understood framework of​​ linear regression. First, we​​​‌ establish negative results showing​ that if an adversary​‌ controls the seed of​​ the generative model, a​​ single synthetic data point​​​‌ can leak as much‌ information as releasing the‌​‌ model itself. Conversely, we​​ show that when synthetic​​​‌ data is generated from‌ random inputs, releasing a‌​‌ limited number of synthetic​​ data points amplifies privacy​​​‌ beyond the model's inherent‌ guarantees. We believe our‌​‌ findings in linear regression​​ can serve as a​​​‌ foundation for deriving more‌ general bounds in the‌​‌ future.

Results: Tighter Privacy​​ Auditing of DP-SGD in​​​‌ the Hidden State Threat‌ Model 46

Participants: Ioan‌​‌ Tudor Cebere, Aurélien​​ Bellet, Nicolas Papernot​​​‌.

Machine learning models‌ can be trained with‌​‌ formal privacy guarantees via​​ differentially private optimizers such​​​‌ as differentially private stochastic‌ gradient descent DP-SGD. In‌​‌ this work, we focus​​ on a threat model​​​‌ where the adversary has‌ access only to the‌​‌ final model, with no​​ visibility into intermediate updates.​​​‌ In the literature, this‌ hidden state threat model‌​‌ exhibits a significant gap​​ between the lower bound​​​‌ from empirical privacy auditing‌ and the theoretical upper‌​‌ bound provided by privacy​​ accounting. To challenge this​​​‌ gap, we propose to‌ audit this threat model‌​‌ with adversaries that craft​​ a gradient sequence designed​​​‌ to maximize the privacy‌ loss of the final‌​‌ model without relying on​​ intermediate updates. Our experiments​​​‌ show that this approach‌ consistently outperforms previous attempts‌​‌ at auditing the hidden​​ state model. Furthermore, our​​​‌ results advance the understanding‌ of achievable privacy guarantees‌​‌ within this threat model.​​ Specifically, when the crafted​​​‌ gradient is inserted at‌ every optimization step, we‌​‌ show that concealing the​​ intermediate model updates in​​​‌ DP-SGD does not enhance‌ the privacy guarantees. The‌​‌ situation is more complex​​ when the crafted gradient​​​‌ is not inserted at‌ every step: our auditing‌​‌ lower bound matches the​​ privacy upper bound only​​​‌ for an adversarially-chosen loss‌ landscape and a sufficiently‌​‌ large batch size. This​​ suggests that existing privacy​​​‌ upper bounds can be‌ improved in certain regimes.‌​‌

Results: When Pattern-by-Pattern Works:​​​‌ Theoretical and Empirical Insights‌ for Logistic Models with‌​‌ Missing Values 65

Participants:​​ Christophe Muller, Julie​​​‌ Josse.

Predicting a‌ response with partially missing‌​‌ inputs remains a challenging​​ task even in parametric​​​‌ models, since parameter estimation‌ in itself is not‌​‌ sufficient to predict on​​ partially observed inputs. Several​​​‌ works study prediction in‌ linear models. In this‌​‌ paper, we focus on​​ logistic models, which present​​​‌ their own difficulties. From‌ a theoretical perspective, we‌​‌ prove that a Pattern-by-Pattern​​ strategy (PbP), which learns​​​‌ one logistic model per‌ missingness pattern, accurately approximates‌​‌ Bayes probabilities in various​​ missing data scenarios such​​​‌ as missing completely at‌ random, missing at random,‌​‌ and missing not at​​ random (MCAR, MAR and​​​‌ MNAR). Empirically, we thoroughly‌ compare various methods (constant‌​‌ and iterative imputations, complete​​ case analysis, PbP, and​​​‌ an EM algorithm) across‌ classification, probability estimation, calibration,‌​‌ and parameter inference. Our​​ analysis provides a comprehensive​​​‌ view on the logistic‌ regression with missing values.‌​‌ It reveals that mean​​ imputation can be used​​​‌ as baseline for low‌ sample sizes, and improved‌​‌ performance is obtained via​​​‌ nonlinear multiple iterative imputation​ techniques with the labels​‌ (MICE.RF.Y). For large sample​​ sizes, PbP is the​​​‌ best method for Gaussian​ mixtures, and we recommend​‌ MICE.RF.Y in presence of​​ nonlinear features.

Results: Do​​​‌ we Need Dozens of​ Methods for Real World​‌ Missing Value Imputation? 65​​

Participants: Krystyna Grzesiak,​​​‌ Christophe Muller, Julie​ Josse, Jeffrey Naef​‌.

Missing values pose​​ a persistent challenge in​​​‌ modern data science. Consequently,​ there is an ever-growing​‌ number of publications introducing​​ new imputation methods in​​​‌ various fields. While many​ studies compare imputation approaches,​‌ they often focus on​​ a limited subset of​​​‌ algorithms and evaluate performance​ primarily through pointwise metrics​‌ such as RMSE, which​​ are not suitable to​​​‌ measure the preservation of​ the true data distribution.​‌ In this work, we​​ provide a systematic benchmarking​​​‌ method based on the​ idea of treating imputation​‌ as a distributional prediction​​ task. We consider a​​​‌ large number of algorithms​ and, for the first​‌ time, evaluate them not​​ only on synthetic missing​​​‌ mechanisms, but also on​ real-world missingness scenarios, using​‌ the concept of Imputation​​ Scores. Finally, while the​​​‌ focus of previous benchmark​ has often been on​‌ numerical data, we also​​ consider mixed data sets​​​‌ in our study. The​ analysis overwhelmingly confirms the​‌ superiority of iterative imputation​​ algorithms, especially the methods​​​‌ implemented in the mice​ R package.

Results: Sodium Bicarbonate​​ for Severe Metabolic Acidemia​​​‌ and Acute Kidney Injury​ 77

Participants: Maxime Fosset​‌, Nicolas Molinari.​​

The effect of sodium​​​‌ bicarbonate infusion on outcome​ in patients with severe​‌ metabolic acidemia and moderate​​ to severe acute kidney​​​‌ injury is unknown. The​ objective is to determine​‌ whether sodium bicarbonate infusion​​ is associated with day​​​‌ 90 all-cause mortality in​ patients with severe metabolic​‌ acidemia and moderate to​​ severe acute kidney injury.​​​‌ Randomized, open-label, clinical trial​ conducted with 640 patients​‌ in 43 French intensive​​ care units from October​​​‌ 6, 2019, to December​ 19, 2023, with 90-day​‌ follow-up. The last date​​ of follow-up was June​​​‌ 17, 2024. Adults with​ severe metabolic acidemia (defined​‌ as pH $\le$7.20)​​ and moderate to severe​​​‌ acute kidney injury were​ enrolled. Intervention Patients were​‌ randomized 1:1 to receive​​ either intravenous sodium bicarbonate​​​‌ infusion or no sodium​ bicarbonate to target an​‌ arterial pH of 7.30​​ or higher.

Main Outcomes​​​‌ and Measures: The primary​ outcome was day 90​‌ all-cause mortality. Secondary outcomes​​ included day 28 and​​​‌ day 180 all-cause mortality;​ use of organ support​‌ therapy, vasopressors, or invasive​​ mechanical ventilation; intensive care​​​‌ unit and hospital length​ of stay; intensive care​‌ unit-acquired infections; fluid balance;​​ day-7 Sequential [Sepsis-related] Organ​​​‌ Failure Assessment score (6​ organ systems' function is​‌ evaluated and scored from​​ 0 [no dysfunction] to​​​‌ 4 [failure]; total score​ ranges from 0 [normal]​‌ to 24 [maximum failure]);​​ and major adverse kidney​​​‌ events on day 90.​ Results Among 640 randomly​‌ assigned patients, 627 were​​ analyzed (313 in the​​​‌ control group and 314​ in the bicarbonate group).​‌ The median age was​​ 67 years (IQR, 59-74​​ years); 194 of 314​​​‌ patients (62$\%$)‌ in the bicarbonate group‌​‌ and 185 of 313​​ controls (59$\%$)​​​‌ were male. In the‌ primary analysis, day 90‌​‌ all-cause mortality was 195​​ of 314 patients (62.1​​​‌$\%$) in the‌ bicarbonate group and 193‌​‌ of 313 (61.7$\%$) in the control​​​‌ group (absolute difference, 0.4;‌ 95$\%$ CI, -7.2‌​‌ to 8.0; P =.91).​​ There was no evidence​​​‌ of a group effect‌ on day 28 or‌​‌ day 180 all-cause mortality.​​ Among 18 secondary outcomes,​​​‌ kidney replacement therapy was‌ used in 109 of‌​‌ 314 (35$\%$)​​ bicarbonate group patients and​​​‌ 157 of 313 (50‌$\%$) controls (absolute‌​‌ difference, -15.5; 95$\%$ CI, -23.1 to -7.8).​​​‌ No evidence of a‌ group effect was found‌​‌ on other secondary outcomes,​​ including adverse events. Conclusions​​​‌ and Relevance: For patients‌ with severe metabolic acidemia‌​‌ and moderate to severe​​ acute kidney injury, intravenous​​​‌ sodium bicarbonate did not‌ affect mortality. Trial Registration‌​‌ ClinicalTrials.gov Identifier: NCT04010630

Results:​​ Allergen Chip Challenge: a​​​‌ nationwide open database supporting‌ allergy prediction algorithms 80‌​‌

Participants: Pascal Demoly.​​

Background: Allergen chip technologies​​​‌ are a powerful tool‌ for simultaneous analysis of‌​‌ hundreds of allergens, generating​​ a comprehensive sensitization landscape​​​‌ for precision medicine in‌ allergy. This considerable amount‌​‌ of data requires extensive​​ knowledge for translation into​​​‌ clinically relevant conclusion. Objective:‌ To harness Machine Learning‌​‌ (ML) for allergen chip​​ interpretation in daily practice,​​​‌ we set out to‌ establish a nationwide, open‌​‌ database of allergen chip,​​ demographic and clinical information​​​‌ and to submit it‌ to an international crowdsourced‌​‌ ML competition to generate​​ a predictive allergy classification​​​‌ algorithm. Methods: The project‌ consortium defined 20 clinical‌​‌ variables and 5 demographic​​ factors for retrospective collection​​​‌ in conjunction with allergen‌ chip IgE data (2014-2023)‌​‌ from 11 French University​​ Hospitals. The dataset was​​​‌ processed to tag confirmed‌ allergy, grade of severity,‌​‌ and culprit allergen identification​​ associated with allergen chip​​​‌ data and submitted to‌ the data challenge. Results:‌​‌ Data were collected for​​ 4,271 patients, yielding a​​​‌ dataset with over 700,000‌ specific IgE data points.‌​‌ Sensitization was present in​​ 3579 patients (84$\%‌$). Allergy was confirmed‌ in 2,236 patients (53‌​‌$\%$) and excluded​​ in 1,076 patients, the​​​‌ remaining 959 being missing‌ outcome data (allergy diagnosis‌​‌ labels). The competition attracted​​ 292 data scientists who​​​‌ submitted 3,135 algorithms. The‌ highest F-scores ranged from‌​‌ 0.780 to 0.786. The​​ database was subsequently made​​​‌ available as an open‌ source. Conclusions: We present‌​‌ a nationwide open allergy​​ database designed to enable​​​‌ the development of predictive‌ algorithms. This scalable framework,‌​‌ integrating clinical data with​​ ML techniques paves the​​​‌ way for data-driven allergen‌ chip use and interpretation‌​‌ by allergists.

10.1.1 Visits of​​ international scientists

10.1.2 Visits‌​‌ to international teams

10.2.1 PEPR Digital Health​​

The "PEPR Santé Numérique",​​​‌ launched in June 2023​ as part of the​‌ Plan Innovation Santé 2030,​​ is a major initiative​​​‌ in the "Digital Health"​ acceleration strategy with a​‌ program dedicated to stimulating​​ scientific research in this​​​‌ field.

PreMeDICaL is involved​ in three projects that​‌ have been launched:

• SMATCH​​ "Statistical and AI Methods​​​‌ for the Challenges of​ Modern Clinical Trials in​‌ Digital Health" - Julie​​ Josse , Pascal Demoly​​​‌Mathieu Even
  • New clinical​ trial methods and designs​‌ based on animal-to-human, research-based​​ disease models,
  • Enriching clinical​​​‌ trials with multi-source, multi-dimensional​ ancillary data,
  • Next-generation designs​‌ for clinical evaluation of​​ digital medical devices based​​​‌ on AI algorithms,
  • Regulation,​ feasibility and dissemination of​‌ clinical trials
• Digital Pharmacological​​ Twins "Multi-scale and longitudinal​​​‌ data modeling in pharmacology​: toward digital pharmacological​‌ twins" - Julie Josse​​Jeffrey NaefClement Berenfeld​​​‌
• Secure, safe and fair​ machine learning for healthcare​‌ - Aurélien Bellet

10.2.2​​ PEPR Cybersecurity

PreMeDICaL is​​​‌ involved in project IPoP​ (Interdisciplinary Project on Privacy)​‌ - Aurélien Bellet .​​ The objectives of this​​​‌ project are to study​ the threats on privacy​‌ that have been introduced​​ by these new services,​​​‌ and to conceive theoretical​ and technical privacy-preserving solutions​‌ that are compatible with​​ French and European regulations,​​​‌ that preserve the quality​ of experience of the​‌ users. These solutions will​​ be deployed and assessed,​​​‌ both on the technological​ and legal sides, and​‌ on their societal acceptability.​​ In order to achieve​​​‌ these objectives, we adopt​ an interdisciplinary approach, bringing​‌ together many diverse fields:​​ computer science, technology, engineering,​​​‌ social sciences, economy and​ law.

The project's scientific​‌ program focuses on new​​ forms of personal information​​​‌ collection, on the learning​ of Artificial Intelligence (AI)​‌ models that preserve the​​ confidentiality of personal information​​​‌ used, on data anonymization​ techniques, on securing personal​‌ data management systems, on​​ differential privacy, on personal​​​‌ data legal protection and​ compliance, and all the​‌ associated societal and ethical​​ considerations. This unifying interdisciplinary​​​‌ research program brings together​ internationally recognized research teams​‌ (from universities, engineering schools​​ and institutions) working on​​​‌ privacy, and the French​ Data Protection Authority (CNIL).​‌

This holistic vision of​​ the issues linked to​​​‌ personal data protection will,​ on one hand, let​‌ us propose solutions to​​ the scientific and technological​​​‌ challenges and, on the​ other hand, help us​‌ confront these solutions in​​ many different ways in​​​‌ the context of interdisciplinary​ collaborations, thus leading to​‌ recommendations and proposals in​​ the field of regulations​​​‌ or legal frameworks. This​ comprehensive consideration of all​‌ the issues aims at​​ encouraging the adoption and​​​‌ acceptability of the solutions​ proposed by all stakeholders,​‌ legislators, data controllers, data​​ processors, solution designers, developers​​​‌ all the way to​ end-users.

10.2.3 Inria Challenge​‌ FedMalin

Aurélien Bellet leads​​ FedMalin. FedMalin is​​​‌ a research project that​ spans 11 Inria research​‌ teams and aims to​​ push Federated Learning (FL)​​​‌ research and concrete use-cases​ through a multidisciplinary consortium​‌ involving expertise in ML,​​ distributed systems, privacy and​​ security, networks, and medicine.​​​‌ We propose to address‌ a number of challenges‌​‌ that arise when FL​​ is deployed over the​​​‌ Internet, including privacy &‌ fairness, energy consumption, personalization,‌​‌ and location/time dependencies. FedMalin​​ will also contribute to​​​‌ the development of open-source‌ tools for FL experimentation‌​‌ and real-world deployments, and​​ use them for concrete​​​‌ applications in medicine and‌ crowdsensing.

The FedMalin Inria‌​‌ Challenge is supported by​​ Groupe La Poste, sponsor​​​‌ of the Inria Foundation.‌

10.3.1 ANR JCJC PRIDE​​

Aurélien Bellet leads PRIDE​​​‌, a JCJC ANR‌ project on privacy-preserving decentralized‌​‌ machine learning. The goal​​ of PRIDE is to​​​‌ develop theoretical and algorithmic‌ tools that enable differentially-private‌​‌ ML methods operating on​​ decentralized datasets, through three​​​‌ complementary objectives:

• Prove that‌ decentralized learning protocols naturally‌​‌ amplify DP guarantees;
• Propose​​ algorithms at the intersection​​​‌ of decentralized ML and‌ secure multi-party computation;
• Design‌​‌ data-adaptive communication schemes to​​ speed up the convergence​​​‌ on heterogeneous datasets.

UM Envi-H:‌​‌

Participants: Pascal Demoly.​​

initiative by the University​​​‌ of Montpellier. The University‌ of Montpellier, with the‌​‌ support of the Regional​​ Health Agency of Occitanie,​​​‌ is launching an innovative‌ project in the field‌​‌ of environmental health education:​​ the creation of a​​​‌ Small Private Online Course‌ (SPOC) dedicated to environmental‌​‌ health (EH) for primary​​ care. This project is​​​‌ part of Axis 1,‌ "Inform, educate, and train‌​‌ in environmental health," of​​ the Regional Environmental Health​​​‌ Plan for Occitanie (PRSE4‌ Occitanie 2023-2028), which "aims‌​‌ to provide professionals, local​​ authorities, and citizens with​​​‌ the knowledge and skills‌ needed to act on‌​‌ environmental and health issues."​​

In collaboration with the​​​‌ Hérault Primary Health Insurance‌ Fund and the University‌​‌ Department of General Medicine,​​ this SPOC will be​​​‌ a hybrid training program‌ combining online modules with‌​‌ in-person sessions.

Available from​​ early 2026, it aims​​​‌ to develop EH skills‌ for learners in both‌​‌ continuing and initial education.​​ It is primarily intended​​​‌ for coordinators of coordinated‌ healthcare structures (Territorial Professional‌​‌ Health Communities - CPTS​​ / Multidisciplinary Health Centers​​​‌ - MSP), as well‌ as for students in‌​‌ related fields.

This program​​ will focus on enhancing​​​‌ the EH competencies of‌ participants through a hybrid‌​‌ format combining online and​​ in-person learning.

ComexIA Health​​​‌ Occitanie:

Participants: Nicolas Molinari‌, Clement Berenfeld.‌​‌

Members of the steering​​ committee for the Occitanie​​​‌ region's key challenge "AI‌ for health": preparation‌​‌ of the call for​​​‌ proposals (12 co-financed PhD​ positions), selection of applications,​‌ dossier follow-up, and management​​ of a 1.2M Euros​​​‌ budget.

Ethical Committee University​ Hospital

Participants: Nicolas Molinari​‌, Julie Josse.​​

We are involved in​​​‌ the CSE (comité, scientifique​ et éthique) for the​‌ CHU Montpellier.

Other local​​ Projects the team is​​​‌ part of: Muse, eDOL,​ expos-UM, viA-UM, Fondation One​‌ Science Montpellier.

11.1.1 Scientific events: organization​

• Aurélien Bellet co-organizes the​‌ Federated Learning One World​​ webinar (1100+ registered attendees)​​​‌ since May 2020.
• Mathieu​ Even : organization of​‌ NeurIPS In Paris,​​ a 2-day event that​​​‌ took place at Sorbonne​ Université on the 25th​‌ and 26th of November​​ 2025, providing a local​​​‌ alternative to the NeurIPS​ conference.
• Linus Bleistein​‌ : took part in​​ the organization of Eurips​​​‌ in Copenhagen, a one-week​ event with over 3500​‌ participants.
• Clement Berenfeld and​​ Linus Bleistein : organization​​​‌ of RAHM 2026,​ a one-day workshop taking​‌ place in PariSanté Campus​​ on the 27th of​​​‌ January 2026, gathering experts​ in health machine learning.​‌

11.1.2 Scientific events: selection​​

11.1.3​​ Journal

11.1.4 Invited​​ talks

• Julie Josse :​​​‌ AI, Science and Society,‌ IPP Paris
• Julie Josse‌​‌ : Data Science, Statistics​​ and Visualisation 2025, July​​​‌ 2025, South Africa. (Online).‌
• Julie Josse : Digicore‌​‌ meeting 2025, European​​ research network on oncology​​​‌
• Julie Josse : Online‌ causal inference seminar
• Julie‌​‌ Josse : LMU AI​​ Keynote Series
• Julie Josse​​​‌ : National Institute for‌ Health and Care Excellence‌​‌ (NICEUK)
• Julie Josse :​​ Toulouse School Economics
• Julie​​​‌ Josse : Academie of‌ science, May 2025.
• Julie‌​‌ Josse : Bernouilli Lab​​, Paris
• Julie Josse​​​‌ : CAUSALab at Harvard‌ T.H. Chan Methods series‌​‌
• Aurélien Bellet : Keynote​​ speaker at the Privacy-Preserving​​​‌ Machine Learning workshop at‌ EurIPS 2025
• Aurélien Bellet‌​‌ : Invited talk at​​ the Trustworthy AI Symposium​​​‌ (Paris AI Action Summit)‌
• Aurélien Bellet : Invited‌​‌ talk at Autumn School​​ on Recent Advances in​​​‌ Machine Learning
• Aurélien Bellet‌ : Invited talk at‌​‌ Dagstuhl Seminar "PETs and​​ AI: Privacy Washing and​​​‌ the Need for a‌ PETs Evaluation Framework"
• Aurélien‌​‌ Bellet : Talk in​​ the BIPID Team, UMR​​​‌ INSERN IAME
• Aurélien Bellet‌ : Invited talk in‌​‌ Bureau de Biostatistique et​​ d'Epidémiologie, ONCOSTAT, Gustave Roussy​​​‌
• Mathieu Even : invited‌ talk at IBS in‌​‌ Liège (19/05/2025-21/05/2025)
• Mathieu Even​​ : invited talk at​​​‌ ICSDS in Sevilla
• Mathieu‌ Even : talk in‌​‌ the PEPR SMATCH days​​ 2025 (07/10/2025)
• Mathieu Even​​​‌ : invited talk at‌ the Biostatistic seminar of‌​‌ CIRC (OMS, Lyon, 04/09/2025)​​
• Mathieu Even : invited​​​‌ talk at the Journées‌ de Société d'Informatique de‌​‌ France to present PhD​​ works.
• Ahmed Boughdiri :​​​‌ invited talk at UC‌ Berkeley to present generalization‌​‌ work (24/11/25)
• Ahmed Boughdiri​​ : invited talk at​​​‌ Soda INRIA to present‌ work on meta-analysis (21/10/25)‌​‌
• Ahmed Boughdiri : invited​​ talk in the PEPR​​​‌ SMATCH days 2025 (07/10/2025)‌
• Christian Janos Lebeda :‌​‌ invited talk at University​​ of Toronto (10/01/2025).
• Christian​​​‌ Janos Lebeda : invited‌ talk at BARC -‌​‌ University of Copenhagen (16/04/2025).​​
• Christian Janos Lebeda :​​​‌ Invited talk at IPoP‌ réunion plénière 2025 (10/10/2025).‌​‌
• Ioan Tudor Cebere :​​ Invited talk at Inria​​​‌ Lille (16/10/2025).
• Clement Berenfeld‌ : Invited talk at‌​‌ IDESP seminar (IDESP, Montpellier,​​ 02/10/2025)
• Clement Berenfeld :​​​‌ Invited talk at the‌ Biostatistic seminar of CIRC‌​‌ (OMS, Lyon, 12/10/2025)
• Jean-Baptiste​​ Fermanian : talk at​​​‌ New challenges in high-dimensional‌ statistics, Marseille.
• Jean-Baptiste Fermanian‌​‌ : talk at VITE​​ 2025, Montpellier.
• Jean-Baptiste Fermanian​​​‌ : poster at Neurips‌ 2025, San Diego.
• Jean-Baptiste‌​‌ Fermanian : talk at​​ UC Berkeley.
• Jean-Baptiste Fermanian​​​‌ : talk at Seminaire‌ Probabilités et Statistiques, Université‌​‌ d'Angers.
• Jean-Baptiste Fermanian :​​ talk at Team seminar​​​‌ MAGNET team, Inria Lille.‌
• Jean-Baptiste Fermanian : talk‌​‌ at MADSTAT Seminar, Toulouse​​ School of Economics, Toulouse.​​​‌
• Laura Fuentes : talk‌ at “Journées de l'IA‌​‌ pour la santé" organized​​ by ANITI and the​​​‌ Occitanie region, Montpellier.
• Laura‌ Fuentes : talk at‌​‌ Seminar#10 IDESP, Montpellier.
• Rémi​​​‌ Khellaf : talk at​ SODA team seminar, October​‌ 2025, Saclay
• Rémi Khellaf​​ : talk at Causal​​​‌ Data Science meeting, October​ 2025, online
• Rémi Khellaf​‌ : talk at Pacific​​ Causal Inference Conference, July​​​‌ 2025, Beijing (online)
• Ghita​ Fassy El Fehri :​‌ talk at ISI 2025,​​ The Hague (05/10/2025-09/10/2025)
• Ghita​​​‌ Fassy El Fehri :​ poster at PPML @​‌ Eurips, Copenhagen (02/12/2025-07/12/2025)

11.1.5​​ Contributed Talks

• Rémi Khellaf​​​‌ : talk at SMATCH​ PEPR annual meeting, November​‌ 2025, Paris
• Rémi Khellaf​​ : talk at Cap25,​​​‌ July 2025, Dijon
• Laura​ Fuentes : talk at​‌ JDS 2025, Marseille.​​
• Rémi Khellaf : talk​​​‌ at Statistics and Biostatistics​ (SNB), November 2025, Paris​‌
• Rémi Khellaf : talk​​ at FedMalin third meeting,​​​‌ November 2025, Montpellier
• Rémi​ Khellaf : talk at​‌ Journées des Statistiques, May​​ 2025, Marseille
• Laura Fuentes​​​‌ : poster at EUROCIM​ 2025 in Ghent (09/04/2025-11/04/2025).​‌
• Ahmed Boughdiri : contributed​​ talk at SNB 2025​​​‌ in Paris (08/10/2025-10/10/2025)
• Mathieu​ Even : contributed talk​‌ at EUROCIM 2025 in​​ Ghent (09/04/2025-11/04/2025)
• Christian Janos​​​‌ Lebeda : 2 contributed​ talks at SOSA 2025​‌ in New Orleans, USA​​ (13/01/2025).
• Christian Janos Lebeda​​​‌ : contributed talk at​ FORC 2025 at Stanford​‌ University, USA (05/06/2025).

11.1.6​​ Leadership within the scientific​​​‌ community

• Julie Josse is​ elected as a member​‌ of the R foundation​​ and of the R​​​‌ Foundation Conference Committee.​ She is in the​‌ board of the French​​ R committee (organization for​​​‌ coordinating R conferences "Les​ rencontres R") and involved​‌ in a task Forwards​​ force on behalf of​​​‌ the R Foundation with​ the aim of increasing​‌ the participation of women​​ and under-represented groups in​​​‌ the STEM community (founding​ member in 2015).
• Charlotte​‌ Voinot : Treasurer of​​ "Groupe Jeunes Statisticien.ne.s"
• Ioan​​​‌ Tudor Cebere : Privacy​ Attacks Workgroup Leadership for​‌ OpenDP.

11.1.7 Scientific expertise​​

1. Aurélien Bellet : Member​​​‌ of the CNIL-Inria Privacy​ Award committee
2. Aurélien Bellet​‌ : Member of the​​ OECD Expert Group on​​​‌ AI, Data, and Privacy​
3. Aurélien Bellet : ethics​‌ advisor for the European​​ Strategy Forum on Research​​​‌ Infrastructures (ESFRI) project SLICES-PP​
4. Aurélien Bellet : Member​‌ of the scientific and​​ ethics committee of AICET​​​‌
5. Clement Berenfeld : member​ of the expert committee​‌ of ANITI IA for​​ health program
6. Julie Josse​​​‌ : Scientific council Cluster​ IA PostGenAI@Paris Sorbonne Université​‌
7. Julie Josse : Evaluation​​ of ERC, and projects​​​‌ for tenured Professor positions​ (Harvard)

11.1.8 Research administration​‌

• Aurélien Bellet : member​​ of the Operational Committee​​​‌ for the assessment of​ Legal and Ethical risks​‌ (COERLE).

11.2.1 Supervision

11.2.2 Juries​​

11.2.3‌ Teaching

• Aurélien Bellet gave‌​‌ a guest lecture for​​ the Responsible Machine Learning​​​‌ course of Master MVA‌ (2.5 hours).
• Clement Berenfeld‌​‌ has taught for the​​ Fondamentaux du machine learning​​​‌ course of Polytech Montpellier‌ (4th year course, 9h).‌​‌
• Julie Josse : Winter​​ School on Causality and​​​‌ Explainable AI (6 hours)‌
• Julie Josse : AI+Science‌​‌ Summer School. University​​ of Chicago-Paris (3 hours).​​​‌
• Julie Josse : Spring‌ School in Statistics and‌​‌ Machine Learning: Oxford-Imperial-Bocconi,​​ 1 week (35 hours).​​​‌

11.3.1 Productions‌ (articles, videos, podcasts, serious‌​‌ games, ...)

• Aurélien Bellet​​ : Participation to the​​​‌ book "Tout comprendre (ou‌ presque) sur l'intelligence artificielle"‌​‌ (Understanding (Almost) Everything About​​​‌ AI), CNRS Editions [link]​.
• Aurélien Bellet :​‌ Interview about AI in​​ participatory democracy in the​​​‌ French media DémocratieS [link]​.
• Aurélien Bellet :​‌ Interview for L'ECO [link]​​
• Mathieu Even was invited​​​‌ by Serge Abiteboul to​ the podcast “Parlez-moi d’IA”​‌ on Cause Commune to​​ talk about his PhD​​​‌ Thesis on federated model​ learning and causal inference​‌ applied to the medical​​ field (link).​​​‌
• Julie Josse : blog​ binaire la recherche: L'intelligence​‌ artificielle et la fabrique​​ du savoir clinique.

11.3.2​​​‌ Participation in Live events​

• Julie Josse : Vivatech​‌, round table Healthcare​​ for All? Balancing Innovation​​​‌ and Equity Using AI.​ video.
• Julie Josse​‌ : Medtech IAvideo​​
• Julie Josse : Presentation​​​‌ de l'équipe Premedical, Mairie​ de Montpellier. Lancement de​‌ de l'association IA Montpellier​​ Méditerrané.

International‌​‌ journals

• 12 articleM.​​Mohamed Akkari, R.​​​‌Régis Lopez, I.‌Isabelle Jaussent, C.‌​‌Célia Vidal, S.​​Sarah Skinner, D.​​​‌Dany Jaffuel and N.‌Nicolas Molinari. Is‌​‌ a change in mouth​​ opening associated with improvements​​​‌ in quality of life‌ in children with type‌​‌ 1 obstructive sleep apnoea​​ after adenotonsillectomy? Protocol for​​​‌ the JawChild prospective cohort‌ study.BMJ Open‌​‌1512December 2025​​, e105324HALDOI​​​‌
• 13 articleA.Arnaud‌ Bourdin, S.Serena‌​‌ Casanova, E.Engi​​ Ahmed, B.Benoit​​​‌ Brouard, A.Adrien‌ Boher, I.Isabelle‌​‌ Vachier, C.Carey​​ Suehs, F.Flore​​​‌ Pujot, E.Erika‌ Nogue, D.David‌​‌ Galeazzi, F.Fanny​​ Cardon, M.Mathilde​​​‌ Volpato, L.Laurence‌ Halimi, S.Sarah‌​‌ Skinner and N.Nicolas​​ Molinari. Larger quality​​​‌ of life gains with‌ asthma support app: a‌​‌ randomised controlled trial.​​ERJ Open ResearchOctober​​​‌ 2025, 00361-2025HAL‌DOI
• 14 articleA.‌​‌Arnaud Bourdin, N.​​Nicolas Malafaye, J.​​​‌Joana Pissarra, E.‌Engi Ahmed, E.‌​‌Erika Nogué, C.​​ M.Carey Meredith Suehs​​​‌, I.Isabelle Vachier‌ and N.Nicolas Molinari‌​‌. Healthcare Resource Utilisation​​ and Costs of Mepolizumab​​​‌ Initiation: A 5‐Year National‌ Cohort Analysis.Allergy‌​‌2025. In press.​​ HALDOI
• 15 article​​​‌Y.Yann Cabon,‌ S.Sebastien Bommart,‌​‌ G.Gregory Marin,​​ E.Engi Ahmed,​​​‌ C.Carey Suehs,‌ I.Isabelle Vachier,‌​‌ A. S.Anne Sophie​​ Gamez, N.Nicolas​​​‌ Molinari and A.Arnaud‌ Bourdin. Uniformity in‌​‌ air trapping distribution is​​ associated with the severity​​​‌ of airway hyper-responsiveness in‌ asthma.European Radiology‌​‌June 2025HALDOI​​
• 16 articleB.Bernard​​​‌ Canaud, A.Andrew‌ Davenport, M.Marion‌​‌ Morena, M.Mailis​​​‌ Amico, N.Nicolas​ Molinari and J.-P.Jean-Paul​‌ Cristol. Continuous beta-2​​ microglobulin–based clearance highlights superiority​​​‌ of high-Dose HDF over​ high-flux HD in predicting​‌ outcomes.Scientific Reports​​151July 2025​​​‌, 20421HALDOI​
• 17 articleM.Mathieu​‌ Capdevila, A.Audrey​​ de Jong, F.​​​‌Fouad Belafia, A.​Aurelie Vonarb, J.​‌Julie Carr, N.​​Nicolas Molinari, O.​​​‌Olivier Choquet, X.​Xavier Capdevila and S.​‌Samir Jaber. Ultrasound​​ guided transcutaneous phrenic nerve​​​‌ stimulation in critically ill​ patients: a new method​‌ to evaluate diaphragmatic function​​.Anesthesiology1423​​​‌2025, 522-531HAL​DOI
• 18 articleM.​‌Mathieu Capdevila, J.​​Joris Pensier, A.​​​‌Audrey de Jong,​ B.Boris Jung,​‌ J.July Beghin,​​ T.Thomas Laumon,​​​‌ Y.Yassir Aarab,​ L.Lucas Deffontis,​‌ T.Thomas Sfara,​​ A.Ambre Cuny,​​​‌ J.Julie Carr,​ N.Nicolas Molinari,​‌ J.-Y.Jean-Yves Le Guennec​​, F.Fabrice Raynaud​​​‌, S.Stefan Matecki​, L.Laurent Brochard​‌, A.Alain Lacampagne​​ and S.Samir Jaber​​​‌. Impact of under-assisted​ ventilation on diaphragm function​‌ and structure in a​​ porcine model.Anesthesiology​​​‌January 2025HALDOI​
• 19 articleM.Matthieu​‌ Conseil, S.Samir​​ Jaber, F.Fabrice​​​‌ Galia, N.Nicolas​ Molinari, G.Gerald​‌ Chanques, A.Audrey​​ de Jong and M.​​​‌Mathieu Capdevila. Neurally​ adjusted ventilatory assist in​‌ critical care patients with​​ and without obesity: a​​​‌ prospective randomized crossover study​.Annals of Intensive​‌ Care151August​​ 2025, 128HAL​​​‌DOI
• 20 articleC.​Camille Desbordes, V.​‌Vivien Szabo, F.​​Frédéric Greco, K.​​​‌Kevin Chalard, C.​Cyril Dargazanli, N.​‌Nicolas Molinari, E.​​Eric Matzner, V.​​​‌Valérie Macioce, J.​Joana Pissarra, G.​‌Gerald Chanques and P.​​ F.Pierre Francois Perrigault​​​‌. Influence of meteorological​ changes on the occurrence​‌ of cerebral aneurysm rupture​​ in the Montpellier region:​​​‌ A retrospective study.​Neurochirurgie712March​‌ 2025, 101630HAL​​DOI
• 21 articleT.​​​‌Tobias Gauss, P.​Pierre Bouzat and J.​‌Julie Josse. From​​ bedside to data and​​​‌ back to bedside: why​ do we need better​‌ data to perform trauma​​ phenotyping and improve trauma​​​‌ care?Intensive Care Medicine​51June 2025,​‌ 1161–1163HALDOI
• 22​​ articleT.Tobias Gauss​​​‌, A.Arthur James​, C.Clelia Colas​‌, N.Nathalie Delhaye​​, M.Mathilde Holleville​​​‌, B.Benjamin Bijok​, M.Marie Werner​‌, A.Alain Meyer​​, V.Véronique Ramonda​​​‌, E.Eric Cesareo​, H.Hugues de​‌ Cherisey, S.Sofiane​​ Medjkoune, S.Samia​​​‌ Salah, J.-P.Jean-Pierre​ Nadal, J.-D.Jean-Denis​‌ Moyer, A.Antoine​​ Vilotitch, P.Pierre​​​‌ Bouzat and J.Julie​ Josse. Comparison of​‌ machine learning and human​​ prediction to identify trauma​​​‌ patients in need of​ hemorrhage control resuscitation (ShockMatrix​‌ study): a prospective observational​​ study.The Lancet​​ Regional Health - Europe​​​‌55August 2025,‌ 101356HALDOI
• 23‌​‌ articleI.-M.Ileana-Maria Ghiordanescu​​, N.Nicolas Molinari​​​‌, A.-M.Ana-Maria Forsea‌, R.Rik Schrijvers‌​‌, C.Cyrille Hoarau​​ and P.P Demoly​​​‌. Penicillin allergy management‌ strategies relevant for clinical‌​‌ practice - a narrative​​ review.Romanian Journal​​​‌ of Internal MedicineJanuary‌ 2025HALDOI
• 24‌​‌ articleT.Timothée Hannequin​​, A.Aurore Ughetto​​​‌, J.Jacob Eliet‌, C.Cinderella Blin‌​‌, A.Aurélien Canon​​, M.Marine Saour​​​‌, P.Philippe Gaudard‌, C.Celia Vidal‌​‌, N.Nicolas Molinari​​, P.Pascal Colson​​​‌ and M.Marc Mourad‌. End-tidal carbon dioxide‌​‌ changes induced by passive​​ leg raising can predict​​​‌ fluid responsiveness in patients‌ on veno-arterial extracorporeal membrane‌​‌ oxygenation: a prospective, interventional​​ study.Annals of​​​‌ Intensive Care151‌November 2025, 185‌​‌HALDOI
• 25 article​​A.Audrey de Jong​​​‌, H.Helena Huguet‌, A.Anne Bignon‌​‌, F.François Stephan​​, T.Thomas Godet​​​‌, L.Lucie Collet‌, K.Karim Asehnoune‌​‌, A.Aude Sylvestre​​, J.Juliette Sautillet​​​‌, R.Raiko Blondonnet‌, M.Martine Ferrandiere‌​‌, Y.Yoann Launey​​, S.Sigismond Lasocki​​​‌, A.Amélie Rollé‌, P.-M.Pierre-Marie Fayolle‌​‌, L.Laurent Muller​​, E.Emmanuel Pardo​​​‌, N.Nicolas Terzi‌, S.Séverin Ramin‌​‌, B.Boris Jung​​, E.Emmanuel Weiss​​​‌, P.-S.Paer-Selim Abback‌, P.Philippe Guerci‌​‌, B.Benjamine Sarton​​, H.Hadrien Rozé​​​‌, C.Claire Dupuis‌, J.Joel Cousson‌​‌, M.Marion Faucher​​, V.Virginie Lemiale​​​‌, B.Bernard Cholley‌, G.Gerald Chanques‌​‌, F.Fouad Belafia​​, C.Claire Chauveton​​​‌, E.Emmanuel Futier‌, E.Elie Azoulay‌​‌, N.Nicolas Molinari​​ and S.Samir Jaber​​​‌. Noninvasive ventilation in‌ postoperative critically ill patients‌​‌ with morbid obesity: secondary​​ analysis of the EXTUBOBESE​​​‌ multicentre randomised clinical trial‌.British Journal of‌​‌ AnaesthesiaOctober 2025HAL​​DOI
• 26 articleL.​​​‌Lucie Knabe, C.‌Camille Tanaka, S.‌​‌Sarah Tebeka, S.​​Samantha Neilson, C.​​​‌Clothilde Cauvin, N.‌Nicolas Mercier, D.‌​‌Daphné Cayron, M.​​Morane Savelli, P.​​​‌Philippe Deruelle, F.‌Florent Fuchs, J.‌​‌Joana Pissarra, N.​​Nicolas Molinari and R.​​​‌Raoul Belzeaux. Prevalence‌ of psychiatric disorders during‌​‌ pregnancy—a feasibility study at​​ second trimester ultrasound in​​​‌ the general population (GROUP‌ study): study protocol.‌​‌BMJ Open152​​2025, e091923HAL​​​‌DOI
• 27 articleC.‌ J.Christian Janos Lebeda‌​‌ and J.Jakub Tetek​​. Better Differentially Private​​​‌ Approximate Histograms and Heavy‌ Hitters using the Misra-Gries‌​‌ Sketch.ACM Transactions​​ on Database Systems50​​​‌3May 2025,‌ 1-26HALDOI
• 28‌​‌ articleA.Amira Marouf​​, S.Sammara Chaubard​​​‌, R.Raphaël Liévin‌, J.-M.Jean-Marie Michot‌​‌, N.Nicolas Molinari​​, J.Julien Rossignol​​​‌, D.Doriane Cavalieri‌, C.Camille Golfier‌​‌, O.Olivier Allangba​​​‌, L.Laure Philippe​, B.Benoît Tessoulin​‌, A.Adrien Chauchet​​, B.Benedicte Deau​​​‌, L.Lucie Obéric​, J.Jacques Vargaftig​‌, A.Aline Moignet​​, A.Aline Clavert​​​‌, R.Rémy Duléry​, G.Gabriel Brisou​‌, S.Stéphanie Tardy​​, V.Virginie Fataccioli​​​‌, R.Roch Houot​, R.-O.René-Olivier Casasnovas​‌, C.Catherine Thieblemont​​, H.Hervé Ghesquieres​​​‌, S.Sylvain Carras​, S.Steven Le​‌ Gouill, G.Guillaume​​ Cartron, V.Vincent​​​‌ Ribrag, M.Morgane​ Cheminant, A.Ambroise​‌ Marçais, F.Felipe​​ Suarez, A.Aurélien​​​‌ Marabelle, O.Olivier​ Tournilhac, G.Gandhi​‌ Damaj, P.Philippe​​ Gaulard, L.Laurence​​​‌ de Leval, F.​François Lemonnier, E.​‌Emmanuel Bachy, S.​​Sylvie Chevret, O.​​​‌Olivier Hermine, L.​Lucile Couronné and A.​‌Arnaud Jaccard. Efficacy​​ of anti‐PD1 therapy in​​​‌ extranodal NK/T cell lymphoma:​ A matched cohort analysis​‌ from the LYSA.​​HemaSphere91January​​​‌ 2025, e70081HAL​DOI
• 29 articleN.​‌Nicolas Molinari, N.​​Nicolas Roche, A.-L.​​​‌Anne-Lise Vataire, S.​Stanislas Perrier, N.​‌Nicolas Pagès, A.​​Arnaud Panes, A.​​​‌Aurélie Schmidt, A.​Arnaud Bourdin and L.​‌Laurence Watier. Economic​​ Burden of Uncontrolled COPD​​​‌ Under Triple Therapy: A​ Population-Based Study.International​‌ Journal of Chronic Obstructive​​ Pulmonary Disease20September​​​‌ 2025, 3295-3304HAL​DOI
• 30 articleC.​‌Clément Monet, L.​​Lauranne Piron, M.​​​‌Marine Pressac, N.​Nicolas Molinari, A.​‌Audrey de Jong,​​ B.Boris Guiu and​​​‌ S.Samir Jaber.​ Study protocol for the​‌ HONIVAH trial: a single-centre​​ randomised study assessing high-flow​​​‌ oxygen therapy versus non-invasive​ ventilation on lung volumes​‌ and the upper airway​​ in hypoxemic critically ill​​​‌ patients.BMJ Open​159September 2025​‌, e106340HALDOI​​
• 31 articleL.Louis​​​‌ Monnier, C.Claude​ Colette, E.Eric​‌ Renard, P.-Y.Pierre-Yves​​ Benhamou, S.Safa​​​‌ Aouinti, N.Nicolas​ Molinari and D.David​‌ Owens. Prevent hypoglycaemia​​ when using automated insulin​​​‌ delivery systems in type​ 1 diabetes requires near​‌ normal glycaemic variability.​​Diabetes & Metabolism51​​​‌1January 2025,​ 101589HALDOI
• 32​‌ articleA.Arthur Naudet-Lasserre​​, J.Joris Pensier​​​‌, A.Audrey de​ Jong, M.Mathieu​‌ Capdevila, S.Samir​​ Jaber and N.Nicolas​​​‌ Molinari. Effect of​ noninvasive ventilation on tracheal​‌ reintubation among patients with​​ hypoxemic respiratory failure following​​​‌ abdominal surgery: a bayesian​ post-hoc analysis of the​‌ NIVAS trial.Critical​​ CareDecember 2025HAL​​​‌DOI
• 33 articleS.​Sam Ng, I.​‌Imad Derraz, G.​​Gaultier Marnat, V.​​​‌Vincent Jecko, J.​Jean Papaxanthos, G.​‌Guillaume Bellanger, C.​​Christophe Cognard, J.-C.​​​‌Jean-Christophe Sol, T.​Tuan Le Van,​‌ M.Moncef Berhouma,​​ P.Pierre Thouant,​​​‌ T.Timothée Jacquesson,​ O.Omer Eker,​‌ H.Hassan El Fertit​​, D.David Liebskind​​, N.Nicolas Molinari​​​‌, N.Nicolas Lonjon‌ and V.Vincent Costalat‌​‌. Embolization of the​​ middle meningeal artery for​​​‌ chronic subdural hematoma: The‌ OTEMACS multicenter, randomized, clinical‌​‌ trial protocol.Interventional​​ NeuroradiologyJuly 2025,​​​‌ 15910199251356098HALDOI
• 34‌ articleJ.Joris Pensier‌​‌, A.Arthur Naudet-Lasserre​​, C.Clément Monet​​​‌, M.Mathieu Capdevila‌, Y.Yassir Aarab‌​‌, I.Ines Lakbar​​, G.Gerald Chanques​​​‌, N.Nicolas Molinari‌, A.Audrey de‌​‌ Jong and S.Samir​​ Jaber. Noninvasive respiratory​​​‌ support following extubation in‌ critically ill adults with‌​‌ obesity: a systematic review​​ and network meta-analysis.​​​‌EClinicalMedicine79January 2025‌, 103002HALDOI‌​‌
• 35 articleJ.Joris​​ Pensier, M.Maha​​​‌ Touaibia, M. A.‌Mohamad Azhar Meerun,‌​‌ P.Philippe Hefteh,​​ N.Nicolas Bloncourt,​​​‌ A.Aurélie Vonarb,‌ A.Albert Prades,‌​‌ B.Bader Al Taweel​​, A.Antoine Debourdeau​​​‌, L.Laurent Monino‌, G.Gérald Chanques‌​‌, N.Nicolas Molinari​​, A.Audrey de​​​‌ Jong, B.Boris‌ Guiu and S.Samir‌​‌ Jaber. Morphological subphenotypes​​ of acute pancreatitis-related acute​​​‌ respiratory distress syndrome.‌Critical CareJanuary 2026‌​‌, Online ahead of​​ printHALDOI
• 36​​​‌ articleV.Vincent Puel‌, F.Frédéric Gagnadoux‌​‌, I.Isabelle Godard​​, G.George Papaioannou​​​‌, N.Naima Zarqane‌, S.Sarz Ibrahim‌​‌, N.Nicolas Molinari​​, J.-L.Jean-Louis Pepin​​​‌ and A.Abdelkebir Sabil‌. Obstructive sleep apnoea‌​‌ syndrome and recurrent vasovagal​​ syncope: Insights from a​​​‌ multicentre observational study.‌Archives of cardiovascular diseases‌​‌November 2025. In​​ press. HALDOI
• 37​​​‌ articleH.Herbert Susmann‌, A.Antoine Chambaz‌​‌, J.Julie Josse​​, P.Philippe Aegerter​​​‌, M.Mathias Wargon‌ and E.Emmanuel Bacry‌​‌. Probabilistic Prediction of​​ Arrivals and Hospitalizations in​​​‌ Emergency Departments in Île-de-France‌.International Journal of‌​‌ Medical Informatics195March​​ 2025, 105728HAL​​​‌DOI
• 38 articleH.‌Herbert Susmann, A.‌​‌Antoine Chambaz, J.​​Julie Josse, P.​​​‌Philippe Aegerter, M.‌Mathias Wargon and E.‌​‌Emmanuel Bacry. Probabilistic​​ prediction of arrivals and​​​‌ hospitalizations in emergency departments‌ in Île-de-France.International‌​‌ Journal of Medical Informatics​​195March 2025,​​​‌ 105728HALDOI
• 39‌ articleC.Celia Vidal‌​‌, F.Fanny Bertelli​​, J.-P.Jean-Pierre Mallet​​​‌, R.Raphael Gilson‌, J.-C.Jean-Christian Borel‌​‌, F.Frédéric Gagnadoux​​, S.Sarah Skinner​​​‌, A.Arnaud Bourdin‌, N.Nicolas Molinari‌​‌ and D.Dany Jaffuel​​. Added value of​​​‌ technician intervention to improve‌ mask management for apneic‌​‌ patients treated with long-term​​ CPAP.Sleep and​​​‌ Breathing291February‌ 2025, 117HAL‌​‌DOI
• 40 articleC.​​Celia Vidal, J.-P.​​​‌Jean-Pierre Mallet, S.‌Sarah Skinner, R.‌​‌Raphael Gilson, O.​​Olivier Gaubert, A.​​​‌Arnaud Prigent, F.‌Frédéric Gagnadoux, J.-C.‌​‌Jean-Christian Borel, A.​​Arnaud Bourdin, N.​​​‌Nicolas Molinari and D.‌Dany Jaffuel. Positive‌​‌ Airway Pressure-Related Aerophagia in​​​‌ Obstructive Sleep Apnea: Results​ from the InterfaceVent Real-Life​‌ Study.Journal of​​ Clinical Medicine1418​​​‌September 2025, 6424​HALDOI
• 41 article​‌C.Celia Vidal,​​ J.-P.Jean-Pierre Mallet,​​​‌ S.Sarah Skinner,​ R.Raphael Gilson,​‌ O.Olivier Gaubert,​​ A.Arnaud Prigent,​​​‌ F.Frédéric Gagnadoux,​ A.Arnaud Bourdin,​‌ N.Nicolas Molinari and​​ D.Dany Jaffuel.​​​‌ Concerns arising from the​ calculation of the apnea-hypopnea​‌ index during CPAP-telemonitoring of​​ patients with obstructive sleep​​​‌ apnea.Respiratory Research​261July 2025​‌, 244HALDOI​​
• 42 articleC.Celia​​​‌ Vidal, J.-P.Jean-Pierre​ Mallet, S.Sarah​‌ Skinner, R.Raphael​​ Gilson, O.Olivier​​​‌ Gaubert, A.Arnaud​ Prigent, F.Frédéric​‌ Gagnadoux, A.Arnaud​​ Bourdin, N.Nicolas​​​‌ Molinari and D.Dany​ Jaffuel. The influence​‌ of atmospheric temperature on​​ long-term CPAP-usage in patients​​​‌ with obstructive sleep apnea.​.Sleep Medicine128​‌January 2025, 22-28​​HALDOI
• 43 article​​​‌J.-B.Jean-Baptiste Woillard,​ S.Sébastien Benzekry,​‌ J.Julie Josse,​​ M.Mélanie White-Koning,​​​‌ E.Etienne Chatelut,​ E.Emmanuelle Comets,​‌ F.Florian Lemaitre,​​ B.Bénédicte Franck,​​​‌ M.Matthieu Grégoire,​ F.Françoise Stanke-Labesque,​‌ S.Sarah Zohar,​​ M.Moreno Ursino and​​​‌ C.Christophe Battail.​ Digital Pharmacological Twins: Bridging​‌ Multi-scale Modelling and Artificial​​ Intelligence for Precision Medicine​​​‌ : the DIGPHAT consortium​.Therapies2025HAL​‌DOI
• 44 articleP.​​Pan Zhao, J.​​​‌Julie Josse and S.​Shu Yang. Efficient​‌ and Robust Transfer Learning​​ of Optimal Individualized Treatment​​​‌ Regimes with Right-Censored Survival​ Data.Journal of​‌ Machine Learning Research26​​48February 2025,​​​‌ 1-54HAL

International peer-reviewed​ conferences

• 45 inproceedingsJ.​‌Judith Abécassis, H.​​Houssam Zenati, S.​​​‌Sami Boumaïza, J.​Julie Josse and B.​‌Bertrand Thirion. CO11.2​​ - Explorer les fonctions​​​‌ cognitives dans UK Biobank​ avec une analyse de​‌ médiation causale.EPICLIN​​ 2025 - Conférence francophone​​​‌ d’EPIdémiologie CLINique73Bordeaux,​ FranceMay 2025,​‌ 203025HALDOI
• 46​​ inproceedingsT.Tudor Cebere​​​‌, A.Aurélien Bellet​ and N.Nicolas Papernot​‌. Tighter Privacy Auditing​​ of DP-SGD in the​​​‌ Hidden State Threat Model​.ICLR 2025 -​‌ 13th International Conference on​​ Learning RepresentationsSingapore, Singapore​​​‌2025HALback to​ text
• 47 inproceedingsA.​‌Alexandre Filiot, N.​​Nicolas Dop, O.​​​‌Oussama Tchita, A.​Auriane Riou, R.​‌Rémy Dubois, T.​​Thomas Peeters, D.​​​‌Daria Valter, M.​Marin Scalbert, C.​‌Charlie Saillard, G.​​Geneviève Robin and A.​​​‌Antoine Olivier. Distilling​ Foundation Models for Robust​‌ and Efficient Models in​​ Digital Pathology.MICCAI​​​‌ 2025 - Open Access​ / SpringerlinkMICCAI 2025​‌ - International Conference on​​ Medical Image Computing and​​​‌ Computer Assisted Intervention15966​Lecture Notes in Computer​‌ ScienceDaejeon, South Korea​​SpringerSeptember 2025,​​​‌ 162-172HALDOI
• 48​ inproceedingsR.Rémi Khellaf​‌, A.Aurélien Bellet​​ and J.Julie Josse​​. Federated Causal Inference:​​​‌ Multi-Study ATE Estimation beyond‌ Meta-Analysis.AISTATS 2025‌​‌ - 28th International Conference​​ on Artificial Intelligence and​​​‌ StatisticsMai Khao, Thailand‌2025HAL
• 49 inproceedings‌​‌C. J.Christian Janos​​ Lebeda. Better Gaussian​​​‌ Mechanism using Correlated Noise‌.2025 Symposium on‌​‌ Simplicity in Algorithms (SOSA)​​SIAM Symposium on Simplicity​​​‌ in Algorithms (SOSA25)New‌ Orleans (Louisiana), United States‌​‌Society for Industrial and​​ Applied MathematicsJanuary 2025​​​‌HALDOI
• 50 inproceedings‌C. J.Christian Janos‌​‌ Lebeda, M.Matthew​​ Regehr, G.Gautam​​​‌ Kamath and T.Thomas‌ Steinke. Avoiding Pitfalls‌​‌ for Privacy Accounting of​​ Subsampled Mechanisms Under Composition​​​‌.IEEE XploreSaTML‌ 2025 - IEEE Conference‌​‌ on Secure and Trustworthy​​ Machine Learning2025 IEEE​​​‌ Conference on Secure and‌ Trustworthy Machine Learning (SaTML)‌​‌Copenhagen, DenmarkIEEEMay​​ 2025, 996-1006HAL​​​‌DOI
• 51 inproceedingsC.‌ J.Christian Janos Lebeda‌​‌ and L.Lukas Retschmeier​​. The Correlated Gaussian​​​‌ Sparse Histogram Mechanism.‌6th Symposium on Foundations‌​‌ of Responsible Computing (FORC​​ 2025)Symposium on Foundations​​​‌ of Responsible Computing (FORC‌ 2025)Standford, United States‌​‌June 2025HALDOI​​
• 52 inproceedingsC. J.​​​‌Christian Janos Lebeda and‌ J.Jakub Tětek.‌​‌ Testing Identity of Distributions​​ under Kolmogorov Distance in​​​‌ Polylogarithmic Space.2025‌ Symposium on Simplicity in‌​‌ Algorithms (SOSA)SIAM Symposium​​ on Simplicity in Algorithms​​​‌ (SOSA25)New orleans, USA,‌ United StatesJanuary 2025‌​‌HALDOI
• 53 inproceedings​​C.Clément Pierquin,​​​‌ A.Aurélien Bellet,‌ M.Marc Tommasi and‌​‌ M.Matthieu Boussard.​​ Privacy Amplification Through Synthetic​​​‌ Data: Insights from Linear‌ Regression.ICML 2025‌​‌ - 42nd International Conference​​ on Machine LearningVancouver,​​​‌ Canada2025HALDOI‌back to text
• 54‌​‌ inproceedingsA. S.Ali​​ Shahin Shamsabadi, P.​​​‌Peter Snyder, R.‌Ralph Giles, A.‌​‌Aurélien Bellet and H.​​Hamed Haddadi. Nebula:​​​‌ Efficient, Private and Accurate‌ Histogram Estimation.CCS‌​‌ 2025 - 32nd ACM​​ Conference on Computer and​​​‌ Communications SecurityTapei (Taiwan),‌ Taiwan2025HAL
• 55‌​‌ inproceedingsH.Houssam Zenati​​, J.Judith Abécassis​​​‌, J.Julie Josse‌ and B.Bertrand Thirion‌​‌. Double Debiased Machine​​ Learning for Mediation Analysis​​​‌ with Continuous Treatments.‌Proceedings of Machine Learning‌​‌ ResearchAISTATS - 28th​​ International Conference on Artificial​​​‌ Intelligence and StatisticsPMLR-‌Mai Khao, ThailandMay‌​‌ 2025HAL

Reports &​​ preprints

• 56 miscJ.​​​‌Judith Abécassis, H.‌Houssam Zenati, S.‌​‌Sami Boumaïza, J.​​Julie Josse and B.​​​‌Bertrand Thirion. Causal‌ mediation analysis with one‌​‌ or multiple mediators: a​​ comparative study.May​​​‌ 2025HAL
• 57 misc‌M.-F.Marie-Félicia Beclin,‌​‌ G.Giulia Marchello,​​ J.-B.Jean-Baptiste Woillard and​​​‌ J.Julie Josse.‌ Super learner for tabular‌​‌ synthetic data generation.​​September 2025HAL
• 58​​​‌ miscA.Aurélien Bellet‌, E.Edwige Cyffers‌​‌, D.Davide Frey​​, R.Romaric Gaudel​​​‌, D.Dimitri Lerévérend‌ and F.François Taïani‌​‌. Unified Privacy Guarantees​​ for Decentralized Learning via​​​‌ Matrix Factorization.October‌ 2025HAL
• 59 misc‌​‌C.Clément Berenfeld,​​​‌ A.Ahmed Boughdiri,​ B.Bénédicte Colnet,​‌ W. A.Wouter A.​​ C. van Amsterdam,​​​‌ A.Aurélien Bellet,​ R.Rémi Khellaf,​‌ E.Erwan Scornet and​​ J.Julie Josse.​​​‌ Causal Meta-Analysis: Rethinking the​ Foundations of Evidence-Based Medicine​‌.May 2025HAL​​back to text
• 60​​​‌ miscL.Linus Bleistein​, A.Aurélien Bellet​‌ and J.Julie Josse​​. Optimal Transport with​​​‌ Heterogeneously Missing Data.​May 2025HAL
• 61​‌ miscT.Thomas Boudou​​, B. L.Batiste​​​‌ Le Bars, N.​Nirupam Gupta and A.​‌Aurélien Bellet. Generalization​​ under Byzantine & Poisoning​​​‌ Attacks: Tight Stability Bounds​ in Robust Distributed Learning​‌.2025HALback​​ to text
• 62 misc​​​‌A.Ahmed Boughdiri,​ C.Clément Berenfeld,​‌ J.Julie Josse and​​ E.Erwan Scornet.​​​‌ A Unified Framework for​ the Transportability of Population-Level​‌ Causal Measures.May​​ 2025HALback to​​​‌ text
• 63 miscM.​Mathieu Even and J.​‌Julie Josse. Rethinking​​ the Win Ratio: A​​​‌ Causal Framework for Hierarchical​ Outcome Analysis.April​‌ 2025HALback to​​ text
• 64 miscG.​​​‌Ghita Fassy El Fehri​, A.Aurélien Bellet​‌ and P.Philippe Bastien​​. Differentially Private and​​​‌ Federated Structure Learning in​ Bayesian Networks.December​‌ 2025HAL
• 65 misc​​ K.Krystyna Grzesiak,​​​‌ C.Christophe Muller,​ J.Julie Josse and​‌ J.Jeffrey Näf.​​ Do we Need Dozens​​​‌ of Methods for Real​ World Missing Value Imputation?​‌ November 2025 HAL back​​ to text back to​​​‌ text
• 66 miscR.​Rémi Khellaf, A.​‌Aurélien Bellet and J.​​Julie Josse. Federated​​​‌ Causal Inference from Multi-Site​ Observational Data via Propensity​‌ Score Aggregation.September​​ 2025HALback to​​​‌ text
• 67 miscC.​ J.Christian Janos Lebeda​‌, M.Mathieu Even​​, A.Aurélien Bellet​​​‌ and J.Julie Josse​. Model Agnostic Differentially​‌ Private Causal Inference.​​May 2025HALback​​​‌ to text
• 68 misc​C. J.Christian Janos​‌ Lebeda, A.Aleksandar​​ Nikolov and H.Haohua​​​‌ Tang. Weighted Fourier​ Factorizations: Optimal Gaussian Noise​‌ for Differentially Private Marginal​​ and Product Queries.​​​‌December 2025HAL
• 69​ miscC.Christophe Muller​‌, E.Erwan Scornet​​ and J.Julie Josse​​​‌. When Pattern-by-Pattern Works:​ Theoretical and Empirical Insights​‌ for Logistic Models with​​ Missing Values.September​​​‌ 2025HAL
• 70 misc​ J.Jeffrey Näf,​‌ E.Erwan Scornet and​​ J.Julie Josse.​​​‌ What Is a Good​ Imputation Under MAR Missingness?​‌ March 2025 HAL
• 71​​ miscG.Gregory Ninot​​​‌, E.Emeline Descamps​, G.Ghislaine Achalid​‌, S.Sébastien Abad​​, F.Fabrice Berna​​​‌, C.Christine Belhomme​, F.François Carbonnel​‌, P. M.Patrizia​​ M. Carrieri, P.​​​‌Patricia Dargent-Molina, F.​Frédéric Fiteni, A.-M.​‌Aude-Marie Foucaut, A.​​Alice Guyon, A.​​​‌Arnaud Legout, B.​Béatrice Lognos, N.​‌Nicolas Molinari, J.​​Julien Nizard, M.​​​‌Michel Noguès, F.​François Paille, P.​‌Pierrick Poisbeau, L.​​Lise Rochaix and B.​​Bruno Falissard. The​​​‌ NPIS Model: A Standardized,‌ Consensus-Based Framework for Evaluating‌​‌ Non-Pharmacological Interventions.April​​ 2025HALDOI
• 72​​​‌ miscV.Vincent Roca‌, M.Marc Tommasi‌​‌, P.Paul Andrey​​, A.Aurélien Bellet​​​‌, M.Markus Schirmer‌, H.Hilde Henon‌​‌, L.Laurent Puy​​, J.Julien Ramon​​​‌, G.Grégory Kuchcinski‌, M.Martin Bretzner‌​‌ and R.Renaud Lopes​​. Federated Learning for​​​‌ MRI-based BrainAGE: a multicenter‌ study on post-stroke functional‌​‌ outcome prediction.2025​​HALDOI
• 73 misc​​​‌M.Mohammad Yaghini,‌ T.Tudor Cebere,‌​‌ M.Michael Menart,​​ A.Aurélien Bellet and​​​‌ N.Nicolas Papernot.‌ Private Rate-Constrained Optimization with‌​‌ Applications to Fair Learning​​.June 2025HAL​​​‌

Other scientific publications

• 74‌ inproceedingsT.Théotime Fehr‌​‌ Delude, T.Tobias​​ Gauss, A.Alexandre​​​‌ Kalimouttou, J.Julie‌ Josse, J.Jules‌​‌ Grèze, P.Pierre​​ Bouzat and B.Benjamin​​​‌ Lemasson. FASTDIAG-TC: Outcome‌ Prediction for Traumatic Brain‌​‌ Injuries: Multimodal AI Combining​​ Clinical Data and CT​​​‌ Scans.IABM 2025‌ - 3e édition du‌​‌ Colloque Français d'Intelligence Artificielle​​ en Imagerie BiomédicaleNice,​​​‌ FranceMarch 2025HAL‌