Papers for
clinical trial designers
Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.
Language models struggle to match expert designs in cancer trial setups
TrialDesignBench: Benchmarking Language Models for Phase-I Trial Design Against Clinical Knowledge
Abstract: AI is used in much of the clinical trial pipeline. However, selecting the statistical design of a trial, a crucial step that determines how drug doses are tested on humans and evaluated, has seen almost no AI progress. LLMs appear to be a natural fit given their strengths in biomedical knowledge and statistical reasoning. Nevertheless, there is currently no appropriate way to evaluate such abilities because recent regulatory and methodological shifts, most notably FDA's Project Optimus, have redefined what constitutes a reasonable design and rendered many previous designs obsolete. To lay the groundwork for AI-assisted trial design, we introduce TrialDesignBench, a benchmark of 617 Phase-I/I-II oncology dose-finding trials, each paired with an expert-validated reference design that is consistent with the latest FDA regulatory initiatives. This reference standard is produced through a formalization of the expert design process. We evaluate five frontier LLMs, none of which matches the reference standard. Our ablation experiments trace this performance gap to two main causes, inference of latent, unrecorded design determinants from trial sources (supplying them raises class-wise balanced accuracy by 0.22) and structured decision-tree reasoning (+0.24). TrialDesignBench offers a reproducible, contemporary-standard objective for evaluating and enhancing AI-assisted trial design.
Patient world models struggle to predict drug effects in trials
Does Your Trial Simulator Know What the Drug Does? An Intervention-Blind Test of Interventional Validity on 7,207 Registered Trials
Abstract: Patient world models are increasingly proposed as simulators of clinical trials, and recent surveys rank their capabilities on a ladder from temporal prediction to counterfactual rollouts. Position on that ladder is currently assigned by reading papers rather than by measurement. We turn the counterfactual rung into a measured quantity. From 30,119 completed phase 2/3 trials with results posted to ClinicalTrials.gov we construct 9,996 audited within-trial contrasts from 7,207 trials, and we score a simulator on how much of its accuracy depends on knowing which intervention was tested. The instrument is an intervention-blind reference: the same model, trained on the same trials, with every intervention-identifying field withheld. Interventional skill is the error the full model removes relative to that reference. On a temporal split (878 test contrasts from 655 trials) a registry-trained gradient-boosting simulator removes a fifth of the error of a trivial baseline, but its interventional skill is -0.12 (95% CI -0.21 to -0.05): withholding the intervention makes it better, and no alternative featurisation, stratum, temporal cut or same-molecule retrieval yields an interval above zero. A frontier language model reading only the registered protocol has interventional skill 0.20 [0.08, 0.31]; smaller and open-weight models sit at the limit of detection. Perturbation tests show the large model uses the intervention as such: swapping the arms reverses the sign of its prediction in 86% of contrasts. The blind reference is itself not drug-blind (a language model recovers the withheld drug class 45% of the time from the blind prompt), so interventional skill is a conservative lower bound. A reliability layer shows that ranking by predicted error lowers it but that calibrated error control does not transfer under temporal shift. Benchmark, code, prompts and per-call model outputs are released.
Transformer model guides safer dose choices in early drug trials
Transformer-Based Reinforcement Learning for Phase I Dose Allocation
Abstract: Phase I dose-finding trials must learn from sparse toxicity observations while limiting exposure to overly toxic doses. We develop a single reinforcement-learning policy for trials with different numbers of ordered dose levels. Each dose is represented as a token containing its normalized position, accumulated patient count, observed dose-limiting toxicities (DLTs), and current-dose indicator. A Transformer encoder summarizes the dose tokens, trial-conditioned cross-attention produces a fixed-dimensional trial representation, and a three-action policy chooses de-escalation, staying, or escalation. The final maximum tolerated dose (MTD) decision is deliberately kept outside the learned policy: after the planned sample size, weighted isotonic regression and a prespecified safety cap select the dose closest to the target toxicity probability. In the simulations, toxicity-penalized policies reduce DLT burden while unpenalized policies improve selection accuracy at the cost of greater toxicity. These results demonstrate a controllable safety--accuracy trade-off but do not establish superiority over conventional dose-finding designs.
Automated system improves clinical trial eligibility criteria using textual evidence
RECITE: Revising Eligibility Criteria Incorporating Textual Evidence
Abstract: Overly restrictive eligibility criteria contribute to widespread enrollment failure in clinical trials, yet published recommendations for how specific criteria should be broadened remain disconnected from trial design at scale. We introduce RECITE (Revising Eligibility Criteria Incorporating Textual Evidence), a framework for automated discovery, extraction, and implementation of evidence-based eligibility criteria revisions. RECITE uses protocol amendment documents from ClinicalTrials.gov to construct a dataset of 11,913 curated instances across 5,735 trials, each pairing original criteria with documentary evidence and target revisions. We benchmark fifteen LLM configurations under truncated-to-context-window and retrieval-augmented generation conditions, on a 3,116-instance sample under an LLM-as-judge protocol validated against ten domain experts (judge vs. human quadratic-weighted Cohen's $κ_{qw} = 0.694 \pm 0.096$). The best open-weight model, Gemma 2 9B, achieves $κ_{qw}$-aligned ordinal score $3.0 \pm 0.9$ on a 0--4 scale (normalized 0.742) and the best closed-weight model, GPT-4o, reaches $3.0 \pm 0.8$ (normalized 0.745). Using an agentic literature discovery system, we then identified 44 paper--trial matches across five publications providing trial-design and EC-modification evidence. Our findings suggest that existing clinical knowledge can be operationalized into actionable trial design recommendations.
Language models struggle to judge scientific evidence but improve with training
Do Language Models Know What Compelling Scientific Evidence Looks Like?
Abstract: Expert scientists excel at weighing empirical findings from multiple sources that vary in both statistical reliability and epistemic relevance. To what extent are language models, which are increasingly driving autonomous discovery systems, capable of such evidence deliberation? We introduce EDB-Bio, a scientific reasoning benchmark curated from real-world data in three biological domains: drug-perturbation transcriptomics, CRISPR screens, and human genome-wide association studies (GWAS). Each of EDB-Bio's 96k episodes presents a sequence of experimental results from published studies and asks the model to predict the probability that a target hypothesis is true. Out-of-the-box, open-weight and frontier models alike demonstrate systematic underconfidence and attain Brier scores comparable to a trivial base-rate forecaster. Post-training against strictly proper scoring rules significantly improves both calibration and discrimination. This calibration pays off downstream: in a sequential experiment selection task, our Ibis0 model is 2.5 times more efficient than the best frontier model. Throughout, quantifying evidence weights as Bayes factors reveals how models update their beliefs over time, whether they agree with expert biologists on relevance, and where they exhibit inconsistencies and biases. Our work extends calibration training and uncertainty analysis to a realistic scientific forecasting setting with applications to human cancer biology. We release all code, data, and trained models.
Algorithm improves best choice finding under budget limits
SHPRR: An Algorithm for Pure Exploration under Constrained Regret
Abstract: Best Arm Identification (BAI) conventionally measures the cost of experimentation by the number of samples collected. In many operational settings, however, the cost of exploration relative to a trusted baseline already in place is equally important. We study BAI under both a fixed sampling horizon and a regret budget. Unlike generic resources considered in constrained pure exploration, regret has a distinctive structure: sampling better arms is less costly relative to the baseline. We theoretically show that the regret cost of exploration decreases across the phases of Sequential Halving, a widely used pure-exploration algorithm. This observation yields improved guarantees on the probability of identifying the best arm and motivates a more effective allocation of the regret budget. We propose Sequential Halving with Proportional Regret Rationing (SHPRR), which front-loads the regret budget to allow more exploration when many arms remain. Experiments on synthetic and real-world instances show that SHPRR improves best-arm identification performance over uniform regret allocation and existing baselines.
Control groups matter in immersive virtual reality therapy trials
Virtual Placebo: a scoping review of control conditions in randomized controlled trials of psychological immersive virtual reality interventions
Abstract: Research on immersive virtual reality (IVR) as a therapeutic tool is growing rapidly, with randomized controlled trials (RCTs) now commonplace. Yet the control conditions in many of these studies that claim strong specific IVR effects vary from usual care to active immersive experiences, making it difficult to draw clinically meaningful conclusions. We conducted a scoping review of recent literature on psychological IVR interventions to assess the main forms of control conditions used and their relations to study outcomes. In this paper, we make the case that most of the recent RCTs studying the effects of specific IVR interventions compare active immersive VR to usual care with no multimedia component. This can lead to strong yet potentially misleading study outcomes and can potentially inflate the IVR interventions' effect. Namely, the lack of an appropriate immersive control condition may risk misattributing general effects of VR devices to the specific content of the intervention. In this review, we discuss the implications of these design choices and propose strategies for including a suitable "virtual placebo" condition to improve the methodological rigor and interpretability of results.
Bayesian method optimizes decision time with error and abstention limits
Bayesian Fixed-Confidence Best-Arm Identification with Abstention
Abstract: We study the Bayesian fixed-confidence best-arm identification problem in which a learner can abstain from making a terminal recommendation. Subject to a target error probability $δ$ and an abstention budget $α$, we propose Screen-and-Confirm (S&C) and analyze its expected stopping time. We further derive the minimum expected stopping time for any algorithm that satisfies these constraints. For Gaussian priors and rewards, as $(δ,α)\downarrow(0,0)$, we establish matching upper and lower bounds up to multiplicative constants, with rate $κ_ν^2[1+\log((α+δ)/δ)]/(α+δ)$, where $κ_ν$ is the prior density of the top-two gap at zero, highlighting that nearly tied instances govern the sampling cost. When abstention is disallowed ($α=0$), S&C yields the optimal $Θ(1/δ)$ dependence for fixed model parameters, removing the additional $\log(1/δ)$ factor in the upper bound of Jang et al. (2024) under our Gaussian assumptions. We further characterize the exact multiplicative constant of the minimum expected stopping time for practical regimes including (i) fixed $α\in (0,1)$ and (ii) $α/δ\to\infty$, which are also achieved by S&C.
Simple method improves picking best items from noisy choices
A Simple and General Approach to top-$k$ Selection
Abstract: This paper considers the Bayesian top-$k$ identification problem, wherein a decision maker is tasked with selecting $k$ items from a pool of items of unknown quality. To assess the items, the decision maker adaptively allocates a fixed sampling budget, where each sample of an item yields a noisy observation of its quality, with the objective of minimizing the expected opportunity cost of the selected set. We propose an approach that allocates each batch of samples by minimizing a surrogate objective derived from an upper bound on the expected opportunity cost. The approach is general: samples can be allocated in batches, the selection size $k$ is arbitrary, and it applies to many sampling distributions. It is also simple and intuitive. We prove that the method's posterior expected opportunity cost decays at an exponential rate of at least half the best rate achievable by any sampling rule. In simulations across a wide range of settings, the method is competitive with, and often improves upon, the best existing algorithm in each setting.
Adaptive sampling method cuts experiments to confirm good designs faster
Halve-Then-Greedy: Adaptive Sampling for Posterior Success with Near-Optimal Instance Dependence
Abstract: We study fixed-budget adaptive experimentation aimed at maximizing the probability of obtaining sufficient posterior evidence that a recommended design's mean performance exceeds a target. Unlike best-arm identification, this objective does not require finding the best design. For Gaussian bandits with a Gaussian prior, we propose \emph{Halve-Then-Greedy} (HTG), which combines a large initial balanced allocation, successive halving, and posterior-greedy sampling. We prove instance-dependent, finite-budget exponential bounds on posterior-failure probability that capture both separation from unfavorable arms and the presence of multiple favorable arms. On one-positive-arm families, we establish an asymptotic sample-complexity lower bound that HTG matches up to a logarithmic factor. We further analyze uniform sampling and sequential halving, revealing regimes where they require more samples than HTG by linear and logarithmic factors, respectively. Simulations illustrate HTG's sample-efficiency advantages and their dependence on instance structure.
Online treatment selection controls errors while learning from limited feedback
Online Individualized Treatment Selection under Partial Feedback: False Discovery Control and Minimax Optimality
Abstract: We study online individualized treatment selection when only the outcome under the assigned treatment is observed. Our objective is to preferentially treat individuals whose potential outcomes under treatment exceed a covariate-specific control reference, while limiting false discoveries among treated individuals. We propose Partial-Feedback Adaptive Individualized Treatment Selection (PAITS), which initializes learning from historical data and combines sequential estimation of local null risks, randomized exploration, and feedback-driven updates of the treatment selection threshold. Under regularity conditions, we establish asymptotic false discovery rate control and sublinear gap-weighted assignment regret relative to a population oracle. We further establish a minimax lower bound on assignment regret under prescribed uniform nuisance-estimation requirements and give conditions under which PAITS attains the corresponding constrained minimax rate up to logarithmic factors. These results quantify the tradeoff between accurate treatment assignment and nuisance learning under partial feedback. Simulation studies and a real-data analysis using the Jester recommendation dataset illustrate the finite-sample performance of PAITS.
Estimation method gauges value of personalized policies in adaptive trials
Inference on the Value of Policy Improvement in Adaptive Experiments
Abstract: In this paper, we develop an asymptotically normal estimator for the value of policy improvement (VPI) under adaptively collected data. This value, which is defined as the gap between the value of the best treatment policy in a class and that of a baseline policy, gauges whether or not it is worth deploying a personalized treatment policy. Building an estimator for the VPI is naturally more difficult than estimating the value of a fixed policy. This is because, in addition to contending with the correlated data and decaying propensities associated with adaptive experiments, any estimator must also handle the non-regularity, or lack of pathwise differentiability, of the underlying estimand when the optimal policy is non-unique. Our proposed estimator works by sequentially fitting a weighted empirical welfare maximizing policy on historical data and then evaluating it using inverse propensity weighting on future samples. We neither assume that the optimal policy is unique nor make a margin assumption on the underlying heterogeneous treatment effects. A key difficulty in analyzing our estimator is showing that the regret incurred through policy learning is asymptotically smaller than the standard error of the estimator. We accomplish this by showing faster regret rates for policy learning under decaying propensities, which may be of independent interest. We not only evaluate the coverage of our estimator on a mix of synthetic and semi-synthetic data, but also employ it to estimate the VPI in an adaptive experiment evaluating the efficacy of job assistance programs in Jordan due to Caria et al. Although our estimates are generally more conservative than their hierarchical Bayesian analysis, which assumes that the optimal policy is unique, we still corroborate the benefits of personalization amongst Syrian refugees.
Cross species neural patterns predict human drug effects in neurological disorders
Cross-species representation learning aligns mouse and human neural dynamics and tracks clinical drug efficacy
Abstract: Preclinical models poorly predict human drug efficacy, particularly in neurological disorders. Neural activity offers a uniquely rich source of translational information because it captures high-dimensional variation in nervous-system function that can be measured in both animals and humans. However, its high dimensionality makes it difficult to distinguish conserved disease-related features from variation arising from species, recording modality and experimental context. Here, we test whether shared neural dynamics can be identified directly from electrophysiology data by learning representations organized by biological state rather than species. We develop a dual-rule contrastive learning framework that aligns corresponding mouse and human states while preserving separation between distinct phenotypes. This framework recovered conserved sensory-response structure across species and, in epilepsy, resolved distinct relationships between three mouse models and heterogeneous human patient populations. When treated animals were projected into a frozen cross-species representation, drug-induced movement towards the human-aligned healthy state retrospectively tracked known clinical efficacy across ten model-drug combinations including a disease-specific detrimental effect. The framework also identified shared disease-associated neural dynamics between Fmr1-knockout mice and human 16p11.2 copy-number variant carriers despite differences in genetic aetiology and recording modality. Together, these findings show the potential of cross-species neural representation learning to map heterogeneous human disease onto experimentally tractable preclinical states and assess whether interventions restore human-relevant circuit function.
LinUCB learns efficiently by sampling like optimal experimental designs
Sampling Allocation of LinUCB: Optimal Design Limits in the Small-Gap Regime
Abstract: We study the sampling allocation of LinUCB in the small-gap regime, where the reward gaps are of order at most $n^{-1/2}$ over the decision horizon $n$. This scaling captures the hard instances underlying worst-case regret lower bounds, for which LinUCB is known to be near optimal up to logarithmic factors in $n$. Using a mean-field perspective, we characterize this allocation through the empirical sampling distribution, a macroscopic object that averages the effect of adaptive decisions over the horizon, and identify its limit as $n\to\infty$. We establish that in this regime, the empirical sampling distribution induced by LinUCB converges to the set of D-optimal designs. This central result reveals that, in the small-gap regime, LinUCB not only achieves near optimal minimax regret but also allocates samples in a way that is asymptotically efficient for learning the reward parameter, thereby connecting regret-driven online learning with information-efficient experimental design. Building on the optimal design limit, we obtain two useful consequences. First, we refine the asymptotic regret analysis of LinUCB in the small-gap regime by characterizing its leading-order constant in the limit. Second, we show that, despite LinUCB's adaptive sampling strategy, the regularized least-squares estimator satisfies a central-limit-type theorem in the small-gap regime, thereby enabling valid statistical inference for the reward parameter.
Clinical trials human AI interactions mapped with new multidimensional framework
Defining and Categorising Human-AI Interactions in Clinical Trials: A Multidimensional Human-AI Classification Approach
Abstract: This paper examines human-AI interactions (HAIIs) in clinical trials and presents a multidimensional categorisation framework that classifies interactions according to AI tasks, human-AI relationships, interaction configurations and interacting human groups. We define HAII, examine existing taxonomies and extend existing categorisation approaches through this novel multidimensional framework. We purposively sampled 15 clinical trials from a previously reported dataset. Each trial was independently categorised by two human reviewers and six large language model (LLM) classifiers. The proposed categorisation provides a structured method for the consistent identification, comparison and synthesis of human-AI interactions across clinical-trial records. The framework is intended to support more consistent comparison and synthesis of AI-related clinical trials and to make explicit the different forms of human involvement associated with AI interventions. The results demonstrate the potential for LLM-assisted categorisation while indicating the continuing importance of human judgement where trial records are incomplete or ambiguous. The principal contribution is a proposed multidimensional framework that brings together AI tasks, human-AI relationships, interaction configurations and interacting human groups within a single approach designed for clinical-trial records. Its significance lies in its potential to support more systematic identification, comparison and synthesis of how humans and AI interact in clinical trials.
Virtual reality designs overlap in cognitive therapy for dementia
Game-Based versus Training-Based Virtual Reality for Cognitive Rehabilitation in Mild Cognitive Impairment and Dementia: A Systematization of Design Paradigms, Outcomes, and Evaluation Rigor
Abstract: Therapeutic exercises can help slow cognitive decline in dementia, but conventional programs are often tedious and difficult to sustain. Virtual reality (VR) offers an alternative through game-based approaches designed to improve engagement and training-based approaches focused on structured functional recovery. Prior reviews report inconsistent findings about these paradigms, while it remains unclear whether the labels represent systems that are technologically distinct. Following PRISMA 2020 guidelines, we systematize 34 studies published between 2015 and 2026, including 16 controlled efficacy trials and 18 design and feasibility studies involving mild cognitive impairment, dementia and Alzheimer's disease. We develop a taxonomy covering design paradigm, immersion, feedback modality, difficulty adaptation and hardware. Our analysis shows substantial overlap between game-based and training-based systems, suggesting that these labels do not clearly distinguish their technological characteristics. We also introduce VR-RES, a seven-dimensional framework for evaluating study quality. Its application highlights recurring limitations in the evidence base, particularly limited follow-up, poor reproducibility and insufficient analysis of demographic factors such as age, sex and education. Based on these findings, we provide recommendations for future VR cognitive rehabilitation research, emphasizing specific design features, rigorous evaluation and improved reporting practices.
NeuronSifter improves planning of CNS interventions by modeling microenvironment dynamics
NeuronSifter: Intervention Planning in CNS Microenvironments
Abstract: Prioritizing central nervous system (CNS) interventions requires predicting how a dose, route, and schedule act on a partially observed microenvironment, then choosing the measurement that would change the decision. Action-conditioned predictors reduce a regimen to an identity token or a scalar exposure, discarding where and when the target is engaged; handing a point estimate to a separate planner then discards the joint uncertainty that makes a measurement worth running. We therefore treat decision quality as a property of the intervention interface, not of controller placement. NeuronSifter compiles regimens into state-conditional target-occupancy fields with support masks, propagates them through microenvironment dynamics with an occupancy-conditioned diffusion operator, and selects measurements by their expected reduction in intervention loss, assimilating typed outcomes into the same posterior. In a declared synthetic Alzheimer's disease (AD) evaluation over 64 paired scenario blocks, occupancy conditioning lowers trajectory continuous ranked probability score from 0.165 to 0.110 and raises intervention ordering accuracy from 0.760 to 0.880, and every paired benchmark contrast remains separated after Holm correction. Decision-directed acquisition attains terminal risk 0.160 against 0.166 for a matched numerical Bayesian experimental design planner, and reaches the target risk at 0.796 $[0.732,0.873]$ of an earlier design control's cost, while the corresponding ratio against the matched planner, 0.963 $[0.907,1.025]$, is not separated from equality; point-state and dependence-ablated interfaces instead raise risk to 0.220 and 0.199, and a full-posterior external controller ties exactly. Published AD trials supply a separate retrospective endpoint bridge.
Minimax policy improves decision making in bernoulli bandit problems
Multi-Armed Bernoulli Bandits via Minimax Single-Arm Stopping
Abstract: We develop an index policy for finite-horizon Bernoulli multi-armed bandits from minimax solutions to single-arm bandit (SAB) problems. Each SAB problem involves choosing between an unknown Bernoulli arm and a known reward. We show that minimizing worst-case regret of SAB problems over all non-anticipative policies admits an exact semi-infinite linear programming formulation. The resulting stopping policies offer a natural way to compare arms: the higher the known reward against which a policy continues sampling, the more promising the unknown arm. We turn this intuition into indices based on cumulative continuation probabilities, with a monotone adjustment and a reward-shortfall cap. By relating index errors to the regret of single-arm stopping policies, we establish a distribution-free regret bound of $4.45\sqrt{KT}+10.75K$ for $K$ arms and horizon $T$. This bound matches the minimax-optimal regret order established in the literature. The guarantee extends to rewards supported on $[0,1]$ through Bernoulli randomization. We also provide a finite-grid implementation with quantified approximation loss. In numerical experiments, the SAB-based index policy achieves lower worst-case regret than every tested benchmark policy across all evaluated numbers of arms and horizons, while closely matching the grid-based MAB minimax policy in the two-arm setting.
AI generated data can change causal experiment questions and results
When AI Generates Covariates: Causal Typing and Estimand Drift in Sequential Experiments
Abstract: AI-generated covariates from notes, conversations, images, and wearable streams can change the causal question when their roles are left unspecified. A generated feature may represent a treatment version, pre-action state, history, design variable, mediator, outcome proxy, observation process, or intercurrent event; these roles are not interchangeable. We formulate a causal type discipline for sequential experiments: a versioned representation map, a causal role classifier, a claim-status filter, and an estimand lock. The lock fixes a standardized proximal effect before generated covariates enter the analysis. Under audit correctness and standard identification assumptions, admissible role assignments preserve this estimand. We apply the established conditional-covariance characterization of compression bias to substitution of generated representations for design-relevant states. A standardized decomposition separates compression, conditional-law, and standardization drift. Further results cover mediator adjustment, post-action leakage, marker-intervention conflation, outcome-guided discovery, and state-measurement error. Cluster-level orthogonal estimators distinguish empirical and superpopulation targets under repeated sessions and missing outcomes. Simulations show that refinement helps when it retains design-relevant information, whereas design erasure, leakage, and same-data marker selection can produce bias or undercoverage. The framework places causal semantics and claim status before confirmatory inference with generated representations.
Evaluation methods guide healthcare large language model testing
A primer on evaluation methods for large language models in healthcare
Abstract: Large language models (LLMs) have a growing range of applications in medicine, and their evaluation is critical for ensuring they provide benefit and not harm. This evaluation can be more challenging than traditional machine learning for many reasons, including probabilistic and open-ended outputs, and behavior that shifts with prompt design and accumulated context. This review covers four key areas of LLM evaluation: principles of study design, statistical methods, capability evaluation and clinical context evaluation. Capability evaluation considers different benchmarks, including multiple-choice, agentic and multi-turn benchmarks, alongside operational metrics like token usage. Clinical context evaluation addresses establishing accuracy of free text outputs, such as human review and LLM-as-a-judge, and clinical trial approaches. Across sections, we describe underlying concepts and potential pitfalls, while emphasizing the importance of aligning evaluation methods with the research question. Together, this article aims to provide a pragmatic basis for designing and executing rigorous evaluations of healthcare LLMs.
Medical studies fall behind fast changing clinical ai models
The widening evaluation gap in medical large language model research 2023 to 2026
Abstract: Large language models are superseded every few quarters; clinical evidence takes years. We asked whether medical research is keeping pace with the systems it evaluates. PubMed returned 11,628 records for January 2023 to June 2026 across fourteen clinical domains, growing 45-fold; 2.5% used a randomised, controlled or prospective design. Evaluation lag, from a study's newest named model release to its own publication, widened from 1.33 to 6.08 quarters. Because discontinued models age mechanically, we benchmarked this against a counterfactual holding model composition fixed: migration to newer systems offset only 56% of the drift (95% CI 50-65). Randomised trials evaluated models a median 4.6 quarters older than other designs (P = 3 x 10^-19), yet among studies naming a model still under development no design differed from any other; 62% of randomised trials evaluated a discontinued family. Rigour and currency are in tension, and that tension reflects model selection rather than research timelines.