252 papers
This paper proposes a causal inference framework that decomposes randomized trial outcomes into biological effects () and preference-mediated choice effects (), arguing that formally identifying both components resolves ambiguity caused by nonadherence and crossover while providing a more complete understanding of treatment efficacy in preference-sensitive care.
This paper introduces a large-scale probabilistic framework that integrates citation structures and semantic trajectories to identify five distinct latent regimes of scientific evolution, demonstrating that semantic drift is the strongest predictor of regime transitions and offering a more nuanced understanding of scientific dynamics beyond traditional citation impact.
This paper revisits the validity of bibliometric indicators under falling production costs by demonstrating that while relative indicators remain robust under uniform cost changes, heterogeneous access requires regime-conditioned reporting using characteristic estimators to avoid misclassification when the underlying distribution shape shifts.
This paper argues that the widely used log(x + 1) transformation introduces substantial bias and misleading inferences, particularly for exponentially distributed data, and recommends abandoning it in favor of alternative numerical methods like Generalized Linear Models (GLM).
This paper proposes and demonstrates a dataset-structured framework for synthesizing machine-learning prediction model evidence that redefines evidence units from publications to validation exercises, thereby correcting misrepresentations of independence and performance that arise from conventional pooling methods.
This study demonstrates that the sharp rise in retraction metadata tagged as computer-generated content from 2010 to 2025 is primarily driven by a few publishers' cleanup programs rather than reflecting the actual prevalence of AI in manuscripts, rendering these counts unsuitable for inferring causal effects of generative AI.
This paper proposes a blockwise joint conformal prediction framework that calibrates prediction sets over operational blocks rather than individual components, demonstrating on hydraulic system data that this approach achieves significantly higher simultaneous coverage for multi-component states compared to conventional marginal methods, albeit with backbone-dependent set sizes and strict reliance on exchangeability assumptions.
This paper describes the systematic development of the AD-DELTA2 checklist, an expert-endorsed extension of the DELTA2 guidance created through a multi-stage process of evidence review, Delphi surveys, and consensus meetings to improve the transparency and reproducibility of sample size reporting in adaptive design trial grant applications and protocols.
This paper presents an exact polynomial-time algorithm that computes the minimum cost of binary modifier-label changes required to alter the sign of an empirical additive treatment-by-modifier interaction, while explicitly clarifying that the method addresses a specific computational instance without establishing broader causal robustness or runtime superiority.
This paper argues that novelty prediction requires explicitly separating category absorption from category opening within a well-specified context, demonstrating that a simple combinatorial null model often outperforms complex rich-get-richer models like the Pitman-Yor process when the task structure is properly accounted for.