arXiv subjects

David Hajage

Publications and source records attributed to David Hajage.

3 recordsLinked to original sources

arXiv2026

Prior-Data Fitted Networks for Causal Inference: a Simulation Study with Real-World Scenarios

Prior-Data Fitted Networks (PFNs) represent a paradigm shift in tabular data prediction. We present the principles of this new paradigm and evaluate two PFNs for estimating the average treatment effect (ATE) of a binary treatment on a binary outcome, using simulated clinical scenarios based on real-world data. We assessed TabPFN combined with causal inference procedures (g-computation and inverse probability of treatment weighting), and CausalPFN, a PFN that directly provides an ATE estimate with a credible interval. Confidence intervals for the TabPFN-based methods were derived using bootstrap resampling. We found that computation times for TabPFN were prohibitive for routine causal inference, particularly because of the need for bootstrapping to yield confidence intervals. Moreover, g-computation with TabPFN produced a highly biased estimator, partially corrected by fitting separate models for each treatment group (T-learner). CausalPFN, by contrast, was computationally efficient but exhibited poor coverage of its 95% credible interval for the ATE, due to both estimation bias and inadequate uncertainty quantification. Beyond automating model specification, some PFN variants - like CausalPFN - attempt to automate causal modeling. In the settings we evaluated, CausalPFN performed poorly. However, new algorithms of this kind continue to be developed, and their application to causal inference tasks requires further investigation.

stat.AP↗

arXiv2025

Assessing the Impact of Covariate Distribution and Positivity Violation on Weighting-Based Indirect Comparisons: a Simulation Study

Population-Adjusted Indirect Comparisons (PAICs) are used to estimate treatment effects when direct comparisons are infeasible and individual patient data (IPD) are only available for one trial. Among PAIC methods, Matching-Adjusted Indirect Comparison (MAIC) is the most widely used. However, little is known about how MAIC performs under challenging conditions such as limited covariate overlap or markedly non-normal covariate distributions. We conducted a Monte Carlo simulation study comparing three estimators: (i) MAIC matching first moment (MAIC-1), (ii) MAIC matching first and second moments (MAIC-2), and (iii) a benchmark method leveraging full IPD -- Propensity Score Weighting (PSW). We examined eight scenarios ranging from ideal conditions to situations with positivity violations and non-normal (including bimodal) covariate distributions. We assessed both anchored and unanchored estimators and examined the impact of adjustment model misspecification. We also applied these estimators to real-world data from the AKIKI and AKIKI-2 trials, comparing renal replacement therapy strategies in critically ill patients. MAIC-1 demonstrated robust performance, remaining unbiased in the presence of moderate positivity violations and non-normal covariates, while MAIC-2 and PSW appeared more sensitive to positivity violations. All methods showed substantial bias when key confounders were omitted, emphasizing the importance of correct model specification. In real-world data, a consistent trend was found with MAIC-1 showing narrower confidence intervals with positivity violation. Our findings support the cautious use of unanchored MAICs and highlight MAIC-1's resilience across moderate violations of assumptions. However, the method's limited flexibility underscores the need for careful use in real-world settings.

stat.ME↗

arXiv2020

Closed-form variance estimators for weighted and stratified dose-response function estimators using generalized propensity score

Propensity score methods are widely used in observational studies for evaluating marginal treatment effects. The generalized propensity score (GPS) is an extension of the propensity score framework, historically developed in the case of binary exposures, for use with quantitative or continuous exposures. In this paper, we proposed variance esti-mators for treatment effect estimators on continuous outcomes. Dose-response functions (DRF) were estimated through weighting on the inverse of the GPS, or using stratification. Variance estimators were evaluated using Monte Carlo simulations. Despite the use of stabilized weights, the variability of the weighted estimator of the DRF was particularly high, and none of the variance estimators (a bootstrap-based estimator, a closed-form estimator especially developped to take into account the estimation step of the GPS, and a sandwich estimator) were able to adequately capture this variability, resulting in coverages below to the nominal value, particularly when the proportion of the variation in the quantitative exposure explained by the covariates was 1 large. The stratified estimator was more stable, and variance estima-tors (a bootstrap-based estimator, a pooled linearized estimator, and a pooled model-based estimator) more efficient at capturing the empirical variability of the parameters of the DRF. The pooled variance estimators tended to overestimate the variance, whereas the bootstrap estimator, which intrinsically takes into account the estimation step of the GPS, resulted in correct variance estimations and coverage rates. These methods were applied to a real data set with the aim of assessing the effect of maternal body mass index on newborn birth weight.

stat.AP↗