arXiv subjects

Adam Kapelner

Publications and source records attributed to Adam Kapelner.

At least 19 recordsLinked to original sources

arXiv2026

Optimal Designs with Robust Inference for Binary Treatment Effects

We study randomized experiments with binary outcomes under Neyman's nonparametric model, where covariate measurements are fixed but potential outcomes are random. In this setting we derive the exact variance of the difference-in-means estimator and characterize designs that minimize it. We show that any balanced design satisfying a covariate-balance condition is asymptotically optimal, and we prove that a broad class of blocking designs satisfies this condition under mild smoothness assumptions. Because the variance depends on unknown success probabilities, unbiased variance estimation is impossible. We therefore develop two conservative estimators: a generalization of the Cochran-Mantel-Haenszel (CMH) statistic applicable to any balanced design, and an extension of Robins' variance estimator for blocking designs. We establish conditions under which the CMH-based estimator is asymptotically tight under local alternatives, thereby yielding asymptotically valid confidence intervals. Our theoretical and simulations results show that blocking and other designs that achieve covariate balance and sufficient degree of randomness, perform well when they are equipped with the CMH-based inference. Thus, we provide an experimental design and inference framework for incidence outcomes that is simultaneously variance-optimal, conservative in finite samples and asymptotically tight.

stat.ME↗

arXiv2026

Predicting Contextual Informativeness for Vocabulary Learning using Deep Learning

We describe a modern deep learning system that automatically identifies informative contextual examples (\qu{contexts}) for first language vocabulary instruction for high school student. Our paper compares three modeling approaches: (i) an unsupervised similarity-based strategy using MPNet's uniformly contextualized embeddings, (ii) a supervised framework built on instruction-aware, fine-tuned Qwen3 embeddings with a nonlinear regression head and (iii) model (ii) plus handcrafted context features. We introduce a novel metric called the Retention Competency Curve to visualize trade-offs between the discarded proportion of good contexts and the \qu{good-to-bad} contexts ratio providing a compact, unified lens on model performance. Model (iii) delivers the most dramatic gains with performance of a good-to-bad ratio of 440 all while only throwing out 70\% of the good contexts. In summary, we demonstrate that a modern embedding model on neural network architecture, when guided by human supervision, results in a low-cost large supply of near-perfect contexts for teaching vocabulary for a variety of target words.

cs.CL↗

arXiv2026

Improved Conditional Logistic Regression using Information in Concordant Pairs with Software

We develop an improvement to conditional logistic regression (CLR) in the setting where the parameter of interest is the additive effect of binary treatment effect on log-odds of the positive level in the binary response. Our improvement is simply to use information learned above the nuisance control covariates found in the concordant response pairs' observations (which is usually discarded) to create an informative prior on their coefficients. This prior is then used in the CLR which is run on the discordant pairs. Our power improvements over CLR are most notable in small sample sizes and in nonlinear log-odds-of-positive-response models. Our methods are released in an optimized R package called bclogit.

stat.ME↗

arXiv2025

Block Designs that Provide Optimal Power in the Cochran-Mantel-Haenszel Test

We consider the asymptotic power performance under local alternatives of the Cochran-Mantel-Haenszel test. Our setting is non-traditional: we investigate randomized experiments that assign subjects via Fisher's blocking design. We show that blocking designs that satisfy a certain balance condition are asymptotically optimal. When the potential outcomes can be ordered, the balance condition is met for all blocking designs with number of blocks going to infinity. More generally, we prove that the pairwise matching design of Greevy et al. (2004) satisfies the balance condition under mild assumptions. In smaller sample sizes, we show a second order effect becomes operational thereby making blocking designs with a smaller number optimal. In practical settings with many covariates, we recommend pairwise matching for its ability to approximate the balance condition.

stat.ME↗

arXiv2024

Development and Validation of a Deep-Learning Model for Differential Treatment Benefit Prediction for Adults with Major Depressive Disorder Deployed in the Artificial Intelligence in Depression Medication Enhancement (AIDME) Study

INTRODUCTION: The pharmacological treatment of Major Depressive Disorder (MDD) relies on a trial-and-error approach. We introduce an artificial intelligence (AI) model aiming to personalize treatment and improve outcomes, which was deployed in the Artificial Intelligence in Depression Medication Enhancement (AIDME) Study. OBJECTIVES: 1) Develop a model capable of predicting probabilities of remission across multiple pharmacological treatments for adults with at least moderate major depression. 2) Validate model predictions and examine them for amplification of harmful biases. METHODS: Data from previous clinical trials of antidepressant medications were standardized into a common framework and included 9,042 adults with moderate to severe major depression. Feature selection retained 25 clinical and demographic variables. Using Bayesian optimization, a deep learning model was trained on the training set, refined using the validation set, and tested once on the held-out test set. RESULTS: In the evaluation on the held-out test set, the model demonstrated achieved an AUC of 0.65. The model outperformed a null model on the test set (p = 0.01). The model demonstrated clinical utility, achieving an absolute improvement in population remission rate in hypothetical and actual improvement testing. While the model did identify one drug (escitalopram) as generally outperforming the other drugs (consistent with the input data), there was otherwise significant variation in drug rankings. On bias testing, the model did not amplify potentially harmful biases. CONCLUSIONS: We demonstrate the first model capable of predicting outcomes for 10 different treatment options for patients with MDD, intended to be used at or near the start of treatment to personalize treatment. The model was put into clinical practice during the AIDME randomized controlled trial whose results are reported separately.

q-bio.NC↗

arXiv2024

The Pairwise Matching Design is Optimal under Extreme Noise and Assignments

We consider the general performance of the difference-in-means estimator in an equally-allocated two-arm randomized experiment under common experimental endpoints such as continuous (regression), incidence, proportion, count and uncensored survival. We consider two sources of randomness: the subject-specific assignments and the contribution of unobserved subject-specific measurements. We then examine mean squared error (MSE) performance under a new, more realistic "simultaneous tail criterion". We prove that the pairwise matching design of Greevy et al. (2004) performs best asymptotically under this criterion when compared to other blocking designs. We also prove that the optimal design must be less random than complete randomization and more random than any deterministic, optimized allocation. Theoretical results are supported by simulations in all five response types.

stat.ME↗

arXiv2023

Towards Outcome-Driven Patient Subgroups: A Machine Learning Analysis Across Six Depression Treatment Studies

Major depressive disorder (MDD) is a heterogeneous condition; multiple underlying neurobiological substrates could be associated with treatment response variability. Understanding the sources of this variability and predicting outcomes has been elusive. Machine learning has shown promise in predicting treatment response in MDD, but one limitation has been the lack of clinical interpretability of machine learning models. We analyzed data from six clinical trials of pharmacological treatment for depression (total n = 5438) using the Differential Prototypes Neural Network (DPNN), a neural network model that derives patient prototypes which can be used to derive treatment-relevant patient clusters while learning to generate probabilities for differential treatment response. A model classifying remission and outputting individual remission probabilities for five first-line monotherapies and three combination treatments was trained using clinical and demographic data. Model validity and clinical utility were measured based on area under the curve (AUC) and expected improvement in sample remission rate with model-guided treatment, respectively. Post-hoc analyses yielded clusters (subgroups) based on patient prototypes learned during training. Prototypes were evaluated for interpretability by assessing differences in feature distributions and treatment-specific outcomes. A 3-prototype model achieved an AUC of 0.66 and an expected absolute improvement in population remission rate compared to the sample remission rate. We identified three treatment-relevant patient clusters which were clinically interpretable. It is possible to produce novel treatment-relevant patient profiles using machine learning models; doing so may improve precision medicine for depression. Note: This model is not currently the subject of any active clinical trials and is not intended for clinical use.

cs.LG↗

arXiv2022

The Optimality of Blocking Designs in Equally and Unequally Allocated Randomized Experiments with General Response

We consider the performance of the difference-in-means estimator in a two-arm randomized experiment under common experimental endpoints such as continuous (regression), incidence, proportion and survival. We examine performance under both equal and unequal allocation to treatment groups and we consider both the Neyman randomization model and the population model. We show that in the Neyman model, where the only source of randomness is the treatment manipulation, there is no free lunch: complete randomization is minimax for the estimator's mean squared error. In the population model, where each subject experiences response noise with zero mean, the optimal design is the deterministic perfect-balance allocation. However, this allocation is generally NP-hard to compute and moreover, depends on unknown response parameters. When considering the tail criterion of Kapelner et al. (2021), we show the optimal design is less random than complete randomization and more random than the deterministic perfect-balance allocation. We prove that Fisher's blocking design provides the asymptotically optimal degree of experimental randomness. Theoretical results are supported by simulations in all considered experimental settings.

math.ST↗

arXiv2022

The Role of Pairwise Matching in Experimental Design for an Incidence Outcome

We consider the problem of evaluating designs for a two-arm randomized experiment with an incidence (binary) outcome under a nonparametric general response model. Our two main results are that the priori pair matching design of Greevy et al. (2004) is (1) the optimal design as measured by mean squared error among all block designs which includes complete randomization. And (2), this pair-matching design is minimax, i.e. it provides the lowest mean squared error under an adversarial response model. Theoretical results are supported by simulations and clinical trial data.

stat.ME↗

arXiv2020

Better Experimental Design by Hybridizing Binary Matching with Imbalance Optimization

We present a new experimental design procedure that divides a set of experimental units into two groups in order to minimize error in estimating an additive treatment effect. One concern is minimizing error at the experimental design stage is large covariate imbalance between the two groups. Another concern is robustness of design to misspecification in response models. We address both concerns in our proposed design: we first place subjects into pairs using optimal nonbipartite matching, making our estimator robust to complicated non-linear response models. Our innovation is to keep the matched pairs extant, take differences of the covariate values within each matched pair and then we use the greedy switching heuristic of Krieger et al. (2019) or rerandomization on these differences. This latter step greatly reduce covariate imbalance to the rate $O_p(n^{-4})$ in the case of one covariate that are uniformly distributed. This rate benefits from the greedy switching heuristic which is $O_p(n^{-3})$ and the rate of matching which is $O_p(n^{-1})$. Further, our resultant designs are shown to be as random as matching which is robust to unobserved covariates. When compared to previous designs, our approach exhibits significant improvement in the mean squared error of the treatment effect estimator when the response model is nonlinear and performs at least as well when it the response model is linear. Our design procedure is found as a method in the open source R package available on CRAN called GreedyExperimentalDesign.

stat.ME↗

arXiv2020

A Matching Procedure for Sequential Experiments that Iteratively Learns which Covariates Improve Power

We propose a dynamic allocation procedure that increases power and efficiency when measuring an average treatment effect in sequential randomized trials exploiting some subjects' previous assessed responses. Subjects arrive sequentially and are either randomized or paired to a previously randomized subject and administered the alternate treatment. The pairing is made via a dynamic matching criterion that iteratively learns which specific covariates are important to the response. We develop estimators for the average treatment effect as well as an exact test. We illustrate our method's increase in efficiency and power over other allocation procedures in both simulated scenarios and a clinical trial dataset. An R package "SeqExpMatch" for use by practitioners is available.

stat.ME↗

arXiv2020

Improving the Power of the Randomization Test

We consider the problem of evaluating designs for a two-arm randomized experiment with the criterion being the power of the randomization test for the one-sided null hypothesis. Our evaluation assumes a response that is linear in one observed covariate, an unobserved component and an additive treatment effect where the only randomness comes from the treatment allocations. It is well-known that the power depends on the allocations' imbalance in the observed covariate and this is the reason for the classic restricted designs such as rerandomization. We show that power is also affected by two other design choices: the number of allocations in the design and the degree of linear dependence among the allocations. We prove that the more allocations, the higher the power and the lower the variability in the power. Designs that feature greater independence of allocations are also shown to have higher performance. Our theoretical findings and extensive simulation studies imply that the designs with the highest power provide thousands of highly independent allocations that each provide nominal imbalance in the observed covariates. These high powered designs exhibit less randomization than complete randomization and more randomization than recently proposed designs based on numerical optimization. Model choices for a practicing experimenter are rerandomization and greedy pair switching, where both outperform complete randomization and numerical optimization. The tradeoff we find also provides a means to specify the imbalance threshold parameter when rerandomizing.

stat.ME↗

arXiv2019

Optimal Rerandomization via a Criterion that Provides Insurance Against Failed Experiments

We present an optimized rerandomization design procedure for a non-sequential treatment-control experiment. Randomized experiments are the gold standard for finding causal effects in nature. But sometimes random assignments result in unequal partitions of the treatment and control group visibly seen as imbalance in observed covariates. There can additionally be imbalance on unobserved covariates. Imbalance in either observed or unobserved covariates increases treatment effect estimator error inflating the width of confidence regions and reducing experimental power. "Rerandomization" is a strategy that omits poor imbalance assignments by limiting imbalance in the observed covariates to a prespecified threshold. However, limiting this threshold too much can increase the risk of contracting error from unobserved covariates. We introduce a criterion that combines observed imbalance while factoring in the risk of inadvertently imbalancing unobserved covariates. We then use this criterion to locate the optimal rerandomization threshold based on the practitioner's level of desired insurance against high estimator error. We demonstrate the gains of our designs in simulation and in a dataset from a large randomized experiment in education. We provide an open source R package available on CRAN named OptimalRerandExpDesigns which generates designs according to our algorithm.

stat.ME↗

arXiv2018

Harmonizing Fully Optimal Designs with Classic Randomization in Fixed Trial Experiments

There is a movement in design of experiments away from the classic randomization put forward by Fisher, Cochran and others to one based on optimization. In fixed-sample trials comparing two groups, measurements of subjects are known in advance and subjects can be divided optimally into two groups based on a criterion of homogeneity or "imbalance" between the two groups. These designs are far from random. This paper seeks to understand the benefits and the costs over classic randomization in the context of different performance criterions such as Efron's worst-case analysis. In the criterion that we motivate, randomization beats optimization. However, the optimal design is shown to lie between these two extremes. Much-needed further work will provide a procedure to find this optimal designs in different scenarios in practice. Until then, it is best to randomize.

stat.ME↗

arXiv2016

Nearly Random Designs with Greatly Improved Balance

We present a new experimental design procedure that divides a set of experimental units into two groups so that the two groups are balanced on a prespecified set of covariates and being almost as random as complete randomization. Under complete randomization, the difference in covariate balance as measured by the standardized average between treatment and control will be $O_p(n^{-1/2})$. If the sample size is not too large this may be material. In this article, we present an algorithm which greedily switches assignment pairs. Resultant designs produce balance of the much lower order $O_p(n^{-3})$ for one covariate. However, our algorithm creates assignments which are, strictly speaking, non-random. We introduce two metrics which capture departures from randomization: one in the style of entropy and one in the style of standard error and demonstrate our assignments are nearly as random as complete randomization in terms of both measures. The results are extended to more than one covariate, simulations are provided to illustrate the results and statistical inference under our design is discussed. We provide an open source R package available on CRAN called GreedyExperimentalDesign which generates designs according to our algorithm.

math.ST↗

arXiv2016

Optimal Experimental Designs for Estimating Henry's Law Constants via the Method of Phase Ratio Variation

When measuring Henry's Law constants ($k_H$) using the phase ratio method via headspace gas chromatography (GC), the value of $k_H$ of the compound under investigation is calculated from the ratio of the slope to the intercept of a linear regression of the the inverse GC response versus the ratio of gas to liquid volumes of a series of vials drawn from the same parent solution. Thus, an experimenter will collect measurements consisting of the independent variable (the gas/liquid volume ratio) and dependent variable (the inverse GC peak area). There is a choice of values of the independent variable during measurement. A review of the literature found that the common approach is a simple uniformly spacing of the liquid volumes. We present an optimal experimental design which estimates $k_H$ with minimum error and provides multiple means for building confidence intervals for such estimates. We illustrate efficiency improvements of our new design with an example measuring the $k_H$ for napthalene in aqueous solution as well as simulations on previous studies. The designs can be easily computed using our open source software optDesignSlopeInt, an R package on CRAN. We also discuss applicability of this method to other fields.

stat.AP↗

arXiv2014

Using Regression Kernels to Forecast A Failure to Appear in Court

Forecasts of prospective criminal behavior have long been an important feature of many criminal justice decisions. There is now substantial evidence that machine learning procedures will classify and forecast at least as well, and typically better, than logistic regression, which has to date dominated conventional practice. However, machine learning procedures are adaptive. They "learn" inductively from training data. As a result, they typically perform best with very large datasets. There is a need, therefore, for forecasting procedures with the promise of machine learning that will perform well with small to moderately-sized datasets. Kernel methods provide precisely that promise. In this paper, we offer an overview of kernel methods in regression settings and compare such a method, regularized with principle components, to stepwise logistic regression. We apply both to a timely and important criminal justice concern: a failure to appear (FTA) at court proceedings following an arraignment. A forecast of an FTA can be an important factor is a judge's decision to release a defendant while awaiting trial and can influence the conditions imposed on that release. Forecasting accuracy matters, and our kernel approach forecasts far more accurately than stepwise logistic regression. The methods developed here are implemented in the R package kernReg currently available on CRAN.

stat.AP↗

arXiv2014

Starvation of Cancer via Induced Ketogenesis and Severe Hypoglycemia

Neoplasms are highly dependent on glucose as their substrate for energy production and are generally not able to catabolize other fuel sources such as ketones and fatty acids. Thus, removing access to glucose has the potential to starve cancer cells and induce apoptosis. Unfortunately, other body tissues are also dependent on glucose for energy under normal conditions. However, in human starvation (or in the setting of diet-induced ketogenesis), the body "keto-adapts" and glucose requirements of most tissues drop to almost nil. Exceptions include the central nervous system (CNS) and various other tissues which have a small but obligatory requirement of glucose. Our hypothesized treatment takes keto-adaptation as a prerequisite. We then propose the induction of severe hypoglycemia by depressing gluconeogenesis while administering glucose to the brain. Although severe hypoglycemia normally produces adverse effects such as seizure and coma, it is relatively safe following ketoadaptation. We hypothesize that our therapeutic hypoglycemia treatment has potential to rapidly induce tumor cell necrosis.

q-bio.OT↗