Research

Restricted mean survival time offers a compelling nonparametric alternative to hazard ratios for right-censored time-to-event data, particularly when the proportional hazards assumption is violated. By capturing the total event-free time over a specified horizon, it provides an intuitive and clinically meaningful measure of absolute treatment benefit. Nonetheless, selecting the restriction time poses challenges: choosing a small restriction time may overlook late-emerging benefits, while a large one can inflate variance and reduce power, an issue whose impact on the precision of inference is often underappreciated. We propose a novel data-driven, adaptive procedure that identifies the optimal restriction time from a continuous range by maximizing a criterion balancing effect size and estimation precision. Consequently, our procedure is particularly powerful when the pattern of the treatment effect is unknown at the design stage. We provide a rigorous theoretical foundation that accounts for the additional variability introduced by adaptive selection. To address nonregular estimation under the null, we develop two complementary strategies: a convex-hull-based estimator, and a penalized approach that regularizes restriction-time selection. Additionally, when restriction time candidates are pre-specified on a discrete grid, our procedure has the same first-order distribution as an oracle estimator evaluated at the penalized population optimizer, with no additional first-order variance from selection. Extensive simulations across realistic survival scenarios demonstrate that our method outperforms traditional restricted mean survival time analyses and the log-rank test, achieving superior power while maintaining approximately nominal type I error rates. In a phase III pancreatic cancer trial with transient treatment effects, our procedure uncovers clinically meaningful benefits that standard methods overlook. Software implementing the methods is available in the package AdaRMST.

Capture–recapture (CRC) surveys are used to estimate the size of a population whose members cannot be enumerated directly. CRC surveys have been used to estimate the number of Coronavirus Disease 2019 (COVID-19) infections, people who use drugs, sex workers, conflict casualties, and trafficking victims. When k-capture samples are obtained, counts of unit captures in subsets of samples are represented naturally by a 2^k contingency table in which one element—the number of individuals appearing in none of the samples—remains unobserved. In the absence of additional assumptions, the population size is not identifiable (i.e., point identified). Stringent assumptions about the dependence between samples are often used to achieve point identification. However, real-world CRC surveys often use convenience samples in which the assumed dependence cannot be guaranteed, and population size estimates under these assumptions may lack empirical credibility. In this work, we apply the theory of partial identification to show that weak assumptions or qualitative knowledge about the nature of dependence between samples can be used to characterize a nontrivial confidence set for the true population size. We construct confidence sets under bounds on pairwise capture probabilities using two methods: test inversion bootstrap confidence intervals and profile likelihood confidence intervals. Simulation results demonstrate well-calibrated confidence sets for each method. In an extensive real-world study, we apply the new methodology to the problem of using heterogeneous survey data to estimate the number of people who inject drugs in Brussels, Belgium.

Genome sequencing technology has generated a vast amount of genomic and epigenomic data, and has provided us a great opportunity to study gene functions on a global scale from an epigenomic view. In the last decade, network-based studies, such as those based on PPI networks and co-expression networks, have shown good performance in capturing functional relationships between genes. However, the functions of a gene and the mechanism of interaction of genes with each other to elucidate their functions are still not entirely clear. Here, we construct a gene co-opening network based on chromatin accessibility of genes. We show that genes related to a specific biological process or the same disease tend to be clustered in the co-opening network. This understanding allows us to detect functional clusters from the network and to predict new functions for genes. We further apply the network to prioritize disease genes for Psoriasis, and demonstrate the power of the joint analysis of the co-opening network and GWAS data in identifying disease genes. Taken together, the co-opening network provides a new viewpoint for the elucidation of gene associations and the interpretation of disease mechanisms.

We present a novel extension of the influential changes-in-changes (CiC) framework of Athey and Imbens (2006) for estimating the average treatment effect on the treated (ATT) and distributional causal effects in panel data with unmeasured confounding. While CiC relaxes the parallel trends assumption in difference-in-differences (DiD), existing methods typically assume a scalar unobserved confounder and monotonic outcome relationships, and lack inference tools that accommodate continuous covariates flexibly. Motivated by empirical settings with complex confounding and rich covariate information, we make two main contributions. First, we establish nonparametric identification under relaxed assumptions that allow high-dimensional, non-monotonic unmeasured confounding. Second, we derive semiparametrically efficient estimators that are Neyman orthogonal to infinite-dimensional nuisance parameters, enabling valid inference even with machine learning-based estimation of nuisance components. We illustrate the utility of our approach in an empirical analysis of mass shootings and U.S. electoral outcomes, where key confounders, such as political mobilization or local gun culture, are typically unobserved and challenging to quantify.

Many real-world processes are trajectories that may be regarded as continuous-time "functional data". Examples include patients' biomarker concentrations, environmental pollutant levels, and prices of stocks. Corresponding advances in data collection have yielded near continuous-time measurements, from e.g. physiological monitors, wearable digital devices, and environmental sensors. Statistical methodology for estimating the causal effect of a time-varying treatment, measured discretely in time, is well developed. But discrete-time methods like the g-formula, structural nested models, and marginal structural models do not generalize easily to continuous time, due to the entanglement of uncountably infinite variables. Moreover, researchers have shown that the choice of discretization time scale can seriously affect the quality of causal inferences about the effects of an intervention. In this paper, we establish causal identification results for continuous-time treatment-outcome relationships for general cadlag stochastic processes under continuous-time confounding, through orthogonalization and weighting. We use three concrete running examples to demonstrate the plausibility of our identification assumptions, as well as their connections to the discrete-time g methods literature.

There are well-established methods for identifying the causal effect of a time-varying treatment applied at discrete time points. However, in the real world, many treatments are continuous or have a finer time scale than the one used for measurement or analysis. While researchers have investigated the discrepancies between estimates under varying discretization scales using simulations and empirical data, it is still unclear how the choice of discretization scale affects causal inference. To address this gap, we present a framework to understand how discretization scales impact the properties of causal inferences about the effect of a time-varying treatment. We introduce the concept of "identification bias", which is the difference between the causal estimand for a continuous-time treatment and the purported estimand of a discretized version of the treatment. We show that this bias can persist even with an infinite number of longitudinal treatment-outcome trajectories. We specifically examine the identification problem in a class of linear stochastic continuous-time data-generating processes and demonstrate the identification bias of the g-formula in this context. Our findings indicate that discretization bias can significantly impact empirical analysis, especially when there are limited repeated measurements. Therefore, we recommend that researchers carefully consider the choice of discretization scale and perform sensitivity analysis to address this bias. We also propose a simple and heuristic quantitative measure for sensitivity concerning discretization and suggest that researchers report this measure along with point and interval estimates in their work. By doing so, researchers can better understand and address the potential impact of discretization bias on causal inference.