Papers

Effective sub-typing (also known as grouping or clustering) of patients using their electronic health record (EHR) data can greatly inform precision medicine efforts. However, subtyping temporal EHR datasets is known to be challenging due to inherent EHR issues, including complexity and irregularity. In this study, we propose a self-supervised Mamba-based model that learns effective EHR representations and enables enhanced patient subtyping. We evaluate the proposed model on public and private real-world EHR datasets to classify the data based on the available labels and subtype patients based on the representations learned from the model. Through an extensive set of experiments, we demonstrate that our model's design choices lead to better performance compared to competitive baseline models for prediction. Moreover, we evaluate several clustering techniques to demonstrate that our findings offer valuable insights into subtyping patients based on temporal records from EHR models\footnote{Our implementations are available at https://github.com/healthylaife/triplet_mamba.

Prediction-powered inference (PPI) refers to a two-level situation where the statistician observes a set of $(x,y)$ pairs and another set of $x$s with the responses $y$ missing. Also available is some independent background data from which a prediction rule $f(x)$ has been produced, perhaps by a machine learning algorithm; $f(x)$ approximates $E\{y\mid x\}$ but there is no guarantee of its accuracy for the situation at hand. Angelopoulos et al. (2023a) developed an algorithm that makes use of all the data, including the unlabeled $x$s, for the estimation of a parameter of interest. A different algorithm is proposed here, using the bootstrap to avoid asymptotics, that is shown to have advantages of efficiency and generality. It is similar in spirit to the original PPI paper by Wang, McCormick and Leek (2020). Prediction-powered inference raises questions about the information available in unlabeled data, with some surprises here, particularly concerning the estimation of the expected value of $y$.

Large language models (LLMs) are increasingly deployed in high-stakes domains, where free-text explanations such as chain-of-thought and post-hoc rationales are used to justify model outputs. Yet it remains unclear whether these explanations are sufficient, i.e., if they contain enough information to explain the model's output-generating process. We generalize classical sufficiency from feature attributions to arbitrary explanations and prove that explanation sufficiency can change depending on the input distribution, which must be explicitly defined for LLM explanations. We propose using the LLM itself to generate alternative inputs conditioned on an explanation, capturing its beliefs about possible inputs. We formalize self-consistent sufficiency as a goal for free-text explanations and introduce an information-theoretic metric, SCSuff, that enables evaluation of free-text explanations without relying on predefined biases or shortcuts. Our experiments show that SCSuff agrees with targeted perturbation tests where applicable and demonstrate that explanation sufficiency can vary with the input distribution. We find LLM explanations are generally insufficient and weakly correlated with model size, accuracy, or output entropy. Analysis of final-token hidden states shows that top and bottom SCSuff scores can be predicted from internal representations, suggesting that SCSuff can guide detection and improvement of sufficient LLM explanations. The code for this paper is available at https://github.com/rajesh-lab/self-consistent-sufficiency .

Prediction sets should have high coverage to be useful, but some coverage notions are more practically relevant than others. In the classification setting, class-conditional coverage requires that the prediction set (i.e., the set of candidate labels for a new test point) must achieve the target accuracy level within each class, which may be challenging to satisfy when many classes are rare and have few calibration points. At the other extreme, marginal coverage requires only that coverage holds on average over the distribution of all classes, which can lead to low-probability labels being essentially ignored. To find a middle ground, recent work has introduced macro-coverage, defined as the unweighted average of class-conditional coverages. Macro-coverage offers a compromise between marginal coverage and class-conditional coverage that is particularly appropriate for long-tailed settings. In this work, we show that label-weighted conformal prediction can be used to produce prediction sets with a finite-sample macro-coverage guarantee, and more generally a guarantee on a family of generalized macro-coverage objectives that aggregate coverage at the level of arbitrary class groupings and take a weighted average. We further characterize the form of the smallest prediction sets satisfying a given generalized macro-coverage objective and propose a corresponding conformal score function. We validate our theoretical results on two large-scale image classification datasets.

Median bias reduction of maximum likelihood estimators can substantially improve estimation and inference. Existing generally applicable methods are, however, implicit, requiring the solution of nonlinear systems of estimating equations for a specified parameterization. Their application to parameter transformations often involves tedious algebra and bespoke implementations. We develop an explicit median bias-corrected estimator for focus parameters that are smooth scalar transformations of a chosen reference parameterization. The estimator results from solving an equation derived from the Cornish-Fisher expansion of the centred and scaled maximum likelihood estimator of the focus parameter, and requires only the ML or an asymptotically equivalent estimator at the reference parameterization, the gradient and Hessian of the transformation, and expectations of products of log-likelihood derivatives. These expectations are available for many models in the bias reduction literature and can also be estimated by Monte Carlo simulation. The resulting estimators are third-order median unbiased and provide one-step approximations to estimators from implicit median bias reduction when the reference parameterization includes the focus parameter. The method can improve standard asymptotic inference and enables hull-based confidence procedures to produce intervals with near nominal finite-sample coverage under median bias control. We illustrate the framework through post-selection inference using the Focused Information Criterion, Mahalanobis distances, quantiles, and scalar focus parameters in regression, stratified, circular, and multiple-mediator models.

Modern machine learning models are trained by optimizing high-dimensional non-convex empirical risk functions. Such cost functions can have a multitude of local optima and yet, gradient-based optimization appears to converge to near-global optima. Within a simple supervised learning setting, we develop a precise picture of which parts of the empirical risk landscape are accessible by polynomial-time algorithms. We are given i.i.d. pairs $\{(\boldsymbol{x}_i,y_i):\; 1 \le i\le n\}$ with $\boldsymbol{x}_i\in \mathbb{R}^d$ standard Gaussian feature vectors, and $y_i\in\mathbb{R}$ response variables that depend on $\boldsymbol{x}_i$ through their projections on an unknown $k$-dimensional subspace. We use empirical risk minimization to learn a model that depends on an $m$-dimensional projection of the data (e.g., an $m$-neurons neural network). We propose an incremental approximate message passing (IAMP) algorithm and precisely characterize the training error it achieves, as well as the relation between test and training error, in the high dimensional asymptotics $n,d\to\infty$, with $n/d\to\alpha \in (0, +\infty)$. Based on earlier work in related models, we expect that the performance achieved by our algorithm is optimal among polynomial-time algorithms.

One of the two dominant approaches for univariate extreme value analysis is to model exceedances above a large threshold, the choice of which has a large impact on inference and whose uncertainty is often subsequently ignored. In this article we review more than 40 threshold selection procedures, including semiparametric methods based on Hill's estimator, visual diagnostics, goodness-of-fit tests, and others based on extended generalized Pareto models. Starting with the statistical properties underlying the various proposals, we provide a critical assessment of their strengths and weaknesses, discuss how they might be automated and describe the results of an extensive simulation study used to identify the most promising procedures. The approaches are compared using a long time series of daily rainfall totals from Padova.

Fractional Brownian motion has been widely used in financial modeling to capture long-range dependence and persistent behavior observed in asset dynamics. In the fractional Black–Scholes framework, accurate estimation of the Hurst parameter is essential, since estimation uncertainty can directly affect option pricing results. In this paper, we propose a Bayesian framework for joint inference on the Hurst parameter and volatility in fractional stochastic differential equation models. In contrast to approaches based solely on point estimation, the proposed method propagates posterior uncertainty directly into option pricing distributions under the fractional Black–Scholes model. Simulation studies are conducted across multiple values of the Hurst parameter and sample sizes to evaluate estimation accuracy, posterior coverage, and pricing uncertainty. The results demonstrate stable posterior inference and coherent uncertainty quantification for both model parameters and option prices. The methodology is further illustrated using WTI crude oil and natural gas data under different market regimes. The empirical analysis indicates that differences in market behavior are driven primarily by changes in volatility rather than strong long-range dependence, while posterior option price distributions reflect substantial variation in pricing uncertainty across regimes. These findings highlight the importance of incorporating joint parameter uncertainty in fractional financial models and demonstrate the practical value of Bayesian methods for option pricing applications.

Classical universal approximation theorems (UAT) establish the expressive power of sigmoidal multilayer perceptrons, but they do not specify how the weights should be initialized. We study a supervised, data-dependent, geometry-aware initialization for one-hidden-layer sigmoidal MLPs that compiles labeled class geometry into network weights. The construction starts from the idea that sigmoid units can act as smooth half-space gates. For each class, we center the training samples at their mean, apply SVD to estimate principal directions and spectral scales, select retained directions by an energy threshold, and represent each retained direction by a pair of sigmoid slab gates. These class-specific gates are then concatenated into a shared hidden layer initialized directly from the training set. We also formulate a SVD-Mahalanobis subspace classifier as a non-neural geometric reference, which tests whether the estimated spectral class geometry is already discriminative before being embedded into the MLP. Experiments on MNIST, Fashion-MNIST, and CIFAR-10 show that the proposed initializer produces a substantially more informative zero-epoch state than a matched task-agnostic Xavier reference, while full training reaches comparable final accuracy. Frozen-hidden experiments and neutral-head ablations further show that the class-wise SVD gates remain useful fixed features, even after the initially aligned output layer is replaced by a Xavier-initialized one.

We study the spectral perturbation of the empirical Fisher Information Matrix (FIM) of a parametric statistical model under two structured perturbations: departure of the input from a reference (in-distribution) ensemble, and finite-precision (quantized) perturbation of the model's parameters. For the first, under an explicit local curvature-monotonicity hypothesis on the dominant eigenvalue lambda_max of the FIM, we show departure from a reference manifold provably elevates lambda_max relative to a calibration baseline (Proposition 3.2), and discuss why this hypothesis is required, since curvature need not increase monotonically under every perturbation. Our principal result is a directional eigenvalue perturbation bound, via Weyl's inequality, showing lambda_max under a quantization noise perturbation is lower bounded by its unperturbed value up to a third-order remainder, and, under a mild genericity condition, strictly exceeds it at leading order (Theorem 4.3). We give two tractable approximations to lambda_max – one heuristic, one with a rigorous two-sided bound – and a completeness result for a threshold-based partition of an augmented state space. These results motivate using sigma_t = lambda_max(F_t)/lambda_base as a runtime monitoring statistic for deployed language models: the quantization result offers a mechanism for an empirical observation of our own, where a calibration threshold for this statistic was approximately 244 times larger than a preliminary full-precision estimate on a 4-bit quantized model, a single measurement rather than a value derived in closed form. We report supporting measurements (twelve models, n=1,080 trajectories) broadly consistent with our predictions, discuss the scope and limitations of every result, and state as an open problem the closed-form prediction of the quantization inflation magnitude our bound does not supply.

Anytime-valid confidence sequences and e-processes are built almost universally from one recipe: average exponential test statistics over a prior on the tilting scale, then invoke Ville's inequality on the resulting nonnegative supermartingale. The mixing prior sets the width of the detection boundary and is usually chosen by hand. We recast the recipe as a two-player game with information as currency. A Learner commits to the prior; Nature adaptively produces a mean-zero score process whose difficulty is priced by a cumulant-generating-function charge. The Learner's mixture wealth obeys a single pathwise Gibbs-variational identity that holds along every realized path with no expectation operator; Ville's inequality, the equalizer condition, the GROW characterization, and the saddlepoint formula are all specializations of it. Three messages organize the rest. First, the law of the iterated logarithm (LIL) is the minimax boundary of this sequential-detection game, not arbitrary combinatorial slack. Second, the optimal prior is not a design choice but the forced equalizer strategy – the unique law that makes every boundary-crossing time equally costly for Nature – and it yields the sharp first iterated-log correction in closed form, with coefficient 3/2 = 1 + 1/2 (one for the Erdős baseline, one half for the Laplace envelope around the saddle). Third, in the log-log scale chart the equalizer is exactly the Jeffreys prior on the scale-of-scales. The Erdős-Kolmogorov integral test is the criterion that selects it. The two-stage finite-time LIL proof, the Howard-Ramdas mixture and stitching constructions, and betting confidence sequences all read as instances of this equalizer principle. A companion empirical evaluation confirms the central identities and locates the Erdős threshold at the predicted value.

Binary classification from positive-only samples is a variant of PAC learning in which the learner receives i.i.d. samples from the positive region of an unknown target concept, but is evaluated under the original distribution (which places mass on both positive and negative regions). This model dates back to Natarajan [1987, STOC], and the characterization of improper learning is well-known – it even appears in textbooks. The characterization of proper positive-only learning, however, has long remained open. In this work, we revisit and settle this question: a concept class is properly learnable from positive-only samples if and only if it has finite VC dimension and satisfies a new combinatorial condition, which we call uniform exterior separability. Together with several separation results, this characterization reveals a surprisingly rich landscape that differs sharply from standard PAC learning: proper and improper learning are separated, randomized and deterministic proper learning are separated, there are classes for which no ERM is a learner, and finite VC dimension does not suffice even for non-uniform learning. Along the way, we introduce new combinatorial dimensions that we believe can be of broader interest in learning theory.

Inference-time scaling is a promising paradigm to improve generative models, especially when outputs must satisfy structural constraints or optimize downstream rewards. We consider Masked Diffusion Model (MDM) and introduce MDM-VGB, a discrete diffusion sampler that augments unmasking generation with theoretically principled reward-guided remasking. Inspired by the recent success of the classical Jerrum-Sinclair backtracking Markov chain in reward-tilted generation, MDM-VGB extends the backtracking random walk from a fixed prefix tree to a masked-state graph, allowing tokens to be unmasked and remasked at arbitrary positions. The resulting sampler favors unmasking and remasking moves that lead to higher-value partial configurations, enabling both effective high-reward generation and efficient repair of low-reward samples. We prove that MDM-VGB is robust to process-verifier noise and achieves quadratic complexity, while popular test-time heuristics such as best-of-$N$ can incur exponential complexity due to error accumulation. Our theoretical findings are corroborated by strong empirical performance, particularly on popular constraint-satisfaction and scientific benchmarks such as Sudoku and QM9.

Background: Both between- and within-subject variations in circadian timing matter for health. If lifestyle changes could be used to regulate circadian timing, they would offer accessible and scalable routes to chronotherapy, but this link remains unclear under real-life conditions. Here, we explore how lifestyle 'traits' (such as typical wake time) and 'states' (day-to-day deviations from traits, such as waking up later than typical) explain between- and within-subject variation in acrophase (peak time) of the circadian rhythm of heart rate (CRHR). Methods: We collected free-living wearable data (smartwatch, continuous glucose monitor) from healthy volunteers for up to 4 weeks. The CRHR was derived from activity-adjusted heart rate, and acrophase was defined as time-of-day at daily CRHR peak. Sleep, food, and physical activity 'factors' were calculated and split into traits and states. Using a linear mixed-effects model, we tested how traits and states associate with between- and within-subject acrophase variance. Findings: Data from 105 healthy volunteers (66 female, age = 42.5 $\pm$ 15.7 years) spanning ~2000 days (18.8 $\pm$ 8.30 days each) were analysed. Traits were substantially more influential than states, explaining 42.3% versus 0.9% of total acrophase variance. Accordingly, traits explained 86.5% of between-subject variance, whereas states explained only 1.8% of within-subject variance. Sleep, food and physical activity factors contributed both jointly and uniquely, and lifestyle timing mattered most. Interpretation: Between-subject lifestyle traits explained acrophase better than within-subject lifestyle states. This asymmetry, alongside the considerable overlap between factors, supports sustained, holistic, timing-focused lifestyle adjustments as chronotherapy targets, testable through future interventional studies.

This paper presents push puppet networks, a novel Bayesian algorithm for structured pruning of large language models. The push puppet network learns a hierarchical function during training that can optimally determine specific network layers to keep for a given target size. By adding a small number of gating parameters via a hierarchical penalty function, the learned smooth function can allow for a network to be resized to very specific sizes without loading the full model into memory or requiring further post-training computation. The method compares favorably with existing approaches (SparseGPT, Wanda) at high pruning sizes (less than 50% of network structure) while realizing measurable speed-ups on conventional GPUs with PyTorch. Furthermore, push puppet networks can achieve significant speedups as candidates for speculative decoding.

Understanding how performance scales jointly with model size and data is a central problem in modern machine learning. Existing theoretical works on scaling laws typically describe generalization as a function of data or compute, often in fixed-feature or infinite-width regimes and for online SGD. Here, we instead study how generalization scales with the number of trainable parameters and the number of samples in a feature-learning model. We analyze $\ell_2$-regularized empirical test error minimization in a quadratic two-layer network in a finite-sample setting with structured data. This setting allows for an explicit characterization of the generalization error as a function of the number of samples, model width, and regularization. Our results reveal a phase diagram with distinct scaling regimes as the number of parameters varies. In particular, the generalization error follows data-dependent power laws controlled by the spectral structure of the target. We further characterize the transitions between regimes, including the onset of interpolation, and their impact on generalization.

Causal representation learning for time series has developed strong identifiability results in discrete-time latent causal models, but identifiability in continuous-time latent stochastic differential equation (SDE) models remains largely open. We address this gap using environment-induced shifts in diffusion covariance. We study additive-noise latent SDEs observed through an unknown nonlinear diffeomorphism, with shared drift but environment-specific diffusion covariance. We show that two diagonal diffusion regimes with pairwise distinct coordinate-wise variance ratios identify the latent coordinates up to permutation, coordinate-wise scaling, and a possible constant shift, without any sparsity assumption on the drift. We first prove this result for linear Ornstein-Uhlenbeck systems and then extend it to general additive-noise latent SDEs. Under mild smoothness, the instantaneous drift-Jacobian causal graph is identifiable up to the same permutation. We propose a two-stage estimator for latent disentanglement and optional graph recovery; experiments on synthetic systems confirm the predicted identifiability boundary, and an application to Hardanger Bridge monitoring data illustrates the approach on real sensor trajectories.

N-of-1 trials, or time-series experiments, are widely used in clinical research and online platforms. Yet the theoretically optimal design for estimating many treatment effects remains unclear. We propose a simple Markovian framework for experimental design in which the treatment assignment process is governed by possibly time-varying transition matrices. This formulation encompasses many existing N-of-1 designs and provides a principled way to control temporal dependence in treatment assignment through Markov transition probabilities. Under a finite-order impulse-response model, we formulate the design objective as minimizing the estimation error of ordinary least squares estimators for target treatment effects, and propose practical design optimization procedures. To characterize the optimal temporal structure, we focus on two structured design classes, random-switch and cycle-switch designs, and establish a complete large-$T$ asymptotic theory for the optimal designs in both classes. Our results justify the robustness of i.i.d. Bernoulli designs in N-of-1 trials and quantify how the optimal design depends on the target estimand, including cumulative and lag-specific treatment effects. Simulations demonstrate the effectiveness and robustness of the proposed designs across multiple scenarios.

Last-iterate convergence and generalization guarantees in first-order convex learning hinge on the monotonicity of the update operator. While linear averaging preserves the monotonicity of gradient updates, this property is often violated when gradients are aggregated non-affinely, as in modern pipelines enforcing constraints like adaptivity, privacy, robustness or fairness. Whether it is possible to design non-affine aggregation rules that maintain monotonicity has remained an open question. We answer this question negatively: we prove that the monotonicity of aggregated gradients is preserved if and only if the aggregation rule is positively affine. Consequently, non-affine aggregation prevents steady convergence and substantially degrade algorithmic stability. We quantify these drawbacks and propose a path forward by identifying sufficient conditions under which monotonicity can be restored. Our results provide a unified theoretical framework explaining the disparate failure modes observed in modern learning systems.

The driveability of a new heavy-truck driveline is traditionally assessed using physical prototypes. Enabling early evaluation of the driving experience in a human-in-the-loop driving simulator using a virtual prototype has the potential to significantly improve development efficiency. To enable driveability assessment using a moving-base simulator, participants must be able to perceive small differences in longitudinal acceleration. The just-noticeable difference (JND) was therefore evaluated for two variants of the classical motion-cueing algorithm (MCA) tuned specifically for tip-in/launch tests and compared to a more general variant in a driving simulator with a long linear track. Psychometric functions were fitted to responses obtained using a weighted staircase procedure and analysed using a generalized linear model. No significant differences in JND were found between the motion cueing variants. The mean JND across all participants and MCA variants was 5.4%. The mean point of subjective equality in the JND experiment was -1.9%, suggesting that participants perceived the acceleration as higher in the second stimulus of a pair. In a subjective comparison, most participants preferred the motion cueing variants that were tuned for launch manoeuvres over the general variant.

The theoretical foundations of the EM algorithm are often thought of in the context of Gaussian mixture models, However, the practical use cases of the EM algorithm span beyond Gaussian models. This paper establishes the first step towards understanding the behavior of the EM algorithm under mixtures of non-Gaussian densities. We show that a mixture of two Exponential distributions can be approximated by the EM algorithm at the sub-Exponential rate of convergence in at most $\log(n)$ iterations. The results here show that extending away from Gaussian mixture models does not affect the statistical performance of the EM algorithm. Furthermore, we present generalizations of typical assumptions in the Gaussian setting like minimum mean-separation and signal-to-noise ratio to the sub-Exponential setting. A simulation study is used to highlight the empirical performance of EM for mixtures of exponentials with promising results for the extension of existing theory to a larger class of mixture models.

Meta-analyses of the accuracy of two diagnostic tests typically assume tests are independent conditional on true disease status. This assumption is often unrealistic and violation leads to biased estimates of the accuracy of tests used in combination. Existing models accounting for conditional dependence require `joint classification' data (results for both tests and the `gold standard' on all participants) from all studies and/or suffer from computational instability. We propose a Bayesian hierarchical model for joint meta-analysis of the accuracy of two binary tests, modelling conditional dependence through study-specific log-odds ratios. The model accommodates studies that do not report joint classification data. We show how the model extends to accommodate data from varied study designs, including studies without a gold standard and studies with partial verification, without assuming imperfect reference standards are error-free. We demonstrate the framework with two example meta-analyses. Our modelling framework retains key features of standard diagnostic test accuracy meta-analysis methods, while allowing for conditional dependence. Ignoring conditional dependence yields biased joint accuracy estimates when conditional dependence is substantial. Our parametrisation maintains computational stability and accommodates data from varied study designs, without requiring an initial data imputation step or assuming error-free reference standards in all studies.

This paper provides a toolkit for the study of distributional treatment effects (DTEs) focused on treatment-effect discontinuities defined as points where marginal distributional effects change sign. Building on the Treatment Effects Curve (TEC, Verme, 2010), the paper makes three contributions. First, we propose a methodological framework comprising a Horizontal Discontinuity Analysis (HDA) comparing groups in regions of opposite-signed effects using causal forests, and a Vertical Discontinuity Analysis (VDA) examining sign-switch points. Second, we adapt crossing-point asymptotics to locate where a TEC crosses zero and to test the non-tangentiality of its local slope with a bias-corrected Wald statistic. Third, we illustrate the full workflow on synthetic data and add a diagnostic application to Mexico's PROGRESA data. The paper shows how these contributions complement and expand existing instruments for DTE analyses.

Quantifying efficacy uncertainty across the entire dose range is crucial in dose-response studies. Although the frequentist simultaneous confidence band (FSCB) is widely used for this purpose, it does not readily incorporate prior knowledge. The Bayesian simultaneous credible band (BSCB) offers a natural alternative, yet practical methods for constructing BSCBs remain scarce in the literature. In this paper, we propose a unified framework for constructing a BSCB for the regression curve in a univariate polynomial model over a finite covariate interval. An efficient simulation-based procedure is developed to determine the critical constant of a BSCB. The framework accommodates inference under different levels of prior information and can be implemented either analytically or via posterior sampling methods. Notably, we prove that under mild regularity conditions, the BSCB is asymptotically equivalent to the FSCB, thereby attaining the nominal frequentist coverage for a broad class of priors. Simulation studies confirm that the BSCB attains the exact posterior simultaneous coverage probability across various scenarios. An application to a dose-response study illustrates its importance in identifying the minimum effective dose in Phase II clinical trials. Software implementation of the proposed methods is available in an accompanying R package.

Benchmarks of machine learning models often include many datasets, making evaluation expensive. For efficiency, it is preferable to perform evaluations on small, representative datasets instead. The selection of such subsets typically relies on heuristics and is rarely analyzed for the robustness of the resulting model rankings. We introduce a framework to perform the task of selecting datasets subsets with an evaluation of how different selection strategies preserve the global model rankings. Our framework includes bootstrap aggregation, which provides valid confidence intervals, allowing a principled comparison of selection strategies. We consider clustering, design criteria (A/D-optimality), random baselines, and greedy farthest-first (FAFI). For the latter, we derive upper bounds on selection quality in terms of ranking errors as a function of the number of selected datasets. Empirically, in time series classification (TSC, 112 datasets) and in a supplementary natural language processing benchmark derived from MTEB (57 tasks), several selection strategies improve rank preservation compared with random subsets, including simple FAFI. In contrast, in recommender systems (30 datasets), the improvement of strategies over random selection is small and typically statistically insignificant. For TSC, our best-performing strategy achieves a Spearman correlation of 0.95 with the full benchmark model rankings using only five selected datasets. Additional experiments indicate that the effectiveness of selection approaches depends on both the quality of dataset representations and the scale of the benchmarking regime.

Background: One of the suggested models for meta-analysis with rare events is the beta-binomial model (BBM). The main advantage of this model compared to inverse-variance models, is that it uses information from zero cells without needing a continuity correction. A disadvantage of the standard BBM is that it ignores randomization. Here we introduce a BBM that respects randomization. Methods: The main idea to preserve randomization is using a common-beta BBM. We illustrate that randomization is preserved by conditioning on the total sum of counts in a studys four-fold table when estimating the model parameters. We perform a simulation study reflecting real-world meta-analyses to compare the models. In addition, we explore in which situations ignoring randomization could be problematic. Results: The BBM that respects randomization performs well in the simulation study that mirrors real meta-analyses in Cochrane and non-Cochrane reviews, respectively. Ignoring randomization appears to be problematic in situations with very different sample sizes of the studies included in the meta-analysis. However, the BBM ignoring randomization tended to perform better when heterogeneity was high. Conclusion: The results show that using the standard BBM, which ignores randomization is usually not biased when the randomization is balanced and the size of studies included in the meta-analysis is not very different. However, as possible ecological bias due to ignoring randomization is an inherent disadvantage of the model and the BBM that respects randomization shows very similar results in the simulation study, it may be generally preferred. Key words Beta-binomial model, generalized linear mixed models, meta-analyses, simulation study, rare events, zero events

Score-based generative models and Langevin samplers rely on estimating the score function $\nabla_x\log p_t(x)$ of a forward diffusion. Classically this is tractable when the drift is linear: the marginal density is Gaussian and the score is a global conditional expectation. For a general nonlinear, state-dependent drift the marginal density has no closed form, and existing methods–denoising score matching and global Fokker–Planck residual penalties–resort to global averaging that inflates estimation error in low-density regions precisely where accuracy is most critical. We address this by developing a local Fokker–Planck geometric framework that replaces global conditioning with local parabolic averaging. Our approach rests on three contributions. First, a time change to the cumulative-variance coordinate reduces the variable-coefficient Fokker–Planck equation to a standard inhomogeneous heat equation, on which we extend Evans' classical heat-ball monotonicity method to derive exact local mean-value representations for the score $\nabla_x\log p$ together with the density, log-density, and entropy density; local well-posedness is established under an explicit dimension-dependent drift budget. Second, for high-dimensional Monte Carlo evaluation of the resulting heat-ball integrals, we introduce the $\kappa$-measure and derive its exact factorized sampler with unit per-sample weight, $\chi^2_2$ radial concentration. Third, the $r\to0$ limit of the heat-ball residual recovers the pointwise Fokker–Planck residual, showing that the local framework is a one-parameter generalization of global FP-residual methods, and that the DSM population minimizer is feasible for the heat-ball constraint at every scale. We validate the framework on 2D structured data on 256-dimensional MNIST, and on a dedicated sampler study confirming the concentration laws.

We derive augmented inverse probability weighted estimators for occupation probabilities of multistate models under two levels of coarsening; right-censoring and baseline exposure. The key exchangeability assumption for identification is coarsening at random, while allowing for time-varying confounders, but not requiring Markov properties. Using existing techniques from causal inference and missing data literature, the derived estimators have highly desirable robustness and efficiency properties. These properties are demonstrated through both theoretical results, and a simulation study.

The widespread collection of fine-grained location data by commercial data brokers creates a re-identification risk that is not widely recognised by the public. While prior research has established that mobility traces are highly unique and that individuals can, in principle, be identified from a handful of spatio-temporal points, such attacks have historically required significant manual effort from skilled analysts, limiting their practical scale. In this feasibility study, we demonstrate in a real world setting that agentic AI fundamentally changes this threat model. We present an end-to-end pipeline in which large language model agents autonomously search the open web, cross-reference public records and social media, and resolve raw coordinate sequences to candidate identities - without human intervention. We evaluate the pipeline on a spatio-temporal dataset containing simulated location points anchored at and around true home and work addresses, focusing on a high-risk disclosure scenario. Our results demonstrate that, from spatio-temporal data and public sources alone, our agentic AI successfully re-identified 18 of the 25 re-identifiable individuals (72%) and 18 of 43 cases overall (41.9%). We discuss implications for Statistical Disclosure Control (SDC) practice and outline the near-future escalation that data custodians and regulators must anticipate. De facto anonymity - an implicit foundation of SDC practice - is shifting. Agentic AI strengthens the case that re-identification is reasonably likely by any means under the GDPR Recital-26 standard, at costs of minutes-and-dollars per target.

Journal recommendation is an important task in scholarly information systems. Existing approaches typically rely on supervised learning models, manually engineered features, or historical interaction data, which may limit their generalizability and interpretability. We propose an LLM-powered semantic alignment framework that formulates journal recommendation as a semantic matching problem between manuscript content and journal scope descriptions. The framework enables large language models (LLMs) to infer journal suitability directly from article titles, abstracts, keywords, and candidate journal information without task-specific training. Experiments are conducted using DeepSeek-V3 on a dataset of 23,609 articles from 49 journals in statistics and related fields. The proposed framework achieves Top-3, Top-5, and Top-10 accuracies of 40.23%, 53.67%, and 70.05%, respectively. Additional analyses show that incorporating reference information generally improves recommendation performance and that recommendations remain highly stable across repeated runs, with an average Top-5 Jaccard similarity of 84%. The framework also generates interpretable reasoning outputs that provide insights into the recommendation process. These findings demonstrate the potential of LLMs as a training-free and scalable paradigm for journal recommendation and scholarly decision support.