arXiv subjects

Iacer Calixto

Publications and source records attributed to Iacer Calixto.

At least 19 recordsLinked to original sources

arXiv2026

Investigating Linear Probe Robustness to Linguistic Register, Medical Specialty, and Corpus Shifts in Medical QA

Linear classifiers trained on hidden states of a large language model (LLM), linear probes, can flag factual errors from a single forward pass. Geometrically, that implies that true and false statements separate along a stable direction in hidden state space, i.e., the truth direction. Prior work disagrees on whether this generalises across input shifts, but the disagreement is hard to interpret because cross-dataset probe transfer experiments confound several kinds of input change at once. We isolate three such variables in medical question-answering (QA): writing style (register), domain (medical specialty), and corpus (dataset). We build a benchmark using 500 MedQA entries, each rewritten into four styles (textbook, patient, clinical note, colloquial), annotated with clinical specialty, and grouped with two other exam corpora, MedMCQA and MMLU-medical, for cross-dataset evaluation. Probing four open-weight LLMs (2--8B), we find that the truth direction is largely robust to writing style (mean $\Delta_\text{register} \approx 0.10$ AUROC on held-out facts) and to medical specialty ($\Delta_\text{specialty} \approx 0.03$), but degrades unevenly across corpora: by $0.12$ AUROC on MMLU-medical and by $0.21$ on MedMCQA, roughly twice the register gap. The register result replicates with a second generator and carries over to human-written patient questions. The truth direction is therefore largely stable within the medical domain but breaks under some corpus shifts, and question format does not explain the break, which suggests that the signal a linear probe recovers is partly bound to dataset structure rather than to medical knowledge alone.

cs.CL↗

arXiv2026

Beyond Scale and Generation: Understanding Language Model-based Entity Matching

Entity matching identifies records that refer to the same real-world entity. Language models can be adapted to this task through bi-encoder, cross-encoder, and generative matcher architectures. However, prior studies often conflate matcher architecture with differences in model backbone, model variant(reflecting different pretraining objectives), and model size, making it difficult to isolate the sources of performance gains. We address this issue through a controlled factorial study spanning three matcher architectures, three model variants and three model sizes from the Qwen3 family, and nine datasets, totaling 1,215 fine-tuning runs. We also evaluate cross-dataset transferability and computational cost. Our results show that model variant is critical for bi-encoders: embedding-oriented variants provide stronger initialization and more favorable representation geometry predictive of downstream matching performance. Cross-encoders retain a consistent advantage over bi-encoders because they jointly encode record pairs rather than representing each record independently, although larger models partially narrow this gap. Generative matchers do not universally outperform cross-encoders. Instead, their advantages concentrate under distribution shift, including subtle unseen differences in record schemas and cross-dataset transfer. We further find that larger models rely more heavily on shortcut learning and therefore do not necessarily perform better. These findings clarify the factors underlying performance differences across matcher architectures and motivate future research and benchmark designs that better disentangle architectural choices from model-level factors while explicitly evaluating distribution shift and cross-dataset transferability. We release our experimental results, code, training scripts, and evaluation data at https://github.com/Jantory/llm-trained-matcher.

cs.DB↗

arXiv2026

Uncertainty Is Not a Safety Net for Clinical VQA, but Can It Anticipate Model Failure?

Safe deployment of clinical vision-language models (VLMs) requires reliable uncertainty estimation (UE): a signal indicating when predictions should be trusted or escalated to a clinician. We test whether current UE methods actually deliver this signal. Benchmarking 8 methods across 12 VLMs on clinical visual question-answering (VQA), we find that UE quality is not an intrinsic property of the UE method: it tracks model accuracy, degrading precisely where the model performance is weakest, and therefore where reliability is most needed. When we stress-test models by hiding the correct option among the multiple-choice answers (NOTA perturbations), accuracy collapses while uncertainty barely changes, leaving models systematically miscalibrated. Yet, we find that uncertainty on the unperturbed input reliably anticipates which predictions will collapse under NOTA, indicating that UE in current VLMs carries diagnostic information about model fragility. Our results position UE as a diagnostic tool for identifying fragile predictions and motivate perturbation-based evaluation as a path toward safe clinical deployment.

cs.CL↗

arXiv2026

Differentially Private De-identification of Dutch Clinical Notes: A Comparative Evaluation

Protecting patient privacy in clinical narratives is essential for enabling secondary use of healthcare data under regulations such as GDPR and HIPAA. While manual de-identification remains the gold standard, it is costly and slow, motivating the need for automated methods that combine privacy guarantees with high utility. Most automated text de-identification pipelines employed named entity recognition (NER) to identify protected entities for redaction. Although methods based on differential privacy (DP) provide formal privacy guarantees, more recently also large language models (LLMs) are increasingly used for text de-identification in the clinical domain. In this work, we present the first comparative study of DP, NER, and LLMs for Dutch clinical text de-identification. We investigate these methods separately as well as hybrid strategies that apply NER or LLM preprocessing prior to DP, and assess performance in terms of privacy leakage and extrinsic evaluation (entity and relation classification). We show that DP mechanisms alone degrade utility substantially, but combining them with linguistic preprocessing, especially LLM-based redaction, significantly improves the privacy-utility trade-off.

cs.CR↗

arXiv2026

Calibrated? Not for Everyone: How Sexual Orientation and Religious Markers Distort LLM Accuracy and Confidence in Medical QA

Safe clinical deployment of Large Language Models (LLMs) requires not only high accuracy but also robust uncertainty calibration to ensure models defer to clinicians when appropriate. Our paper investigates how social descriptors of a patient (specifically sexual orientation and religious affiliation) distort these uncertainty signals and model accuracy. Evaluating nine general-purpose and biomedical LLMs on 2,364 medical questions and their counterfactual variants, we demonstrate that identity markers cause a "calibration crisis". "Homosexual" markers consistently trigger performance drops, and intersectional identities produce idiosyncratic, non-additive harms to calibration. Moreover, a clinician-validated case study in an open-ended generation setting confirms that these failures are not an artifact of the multiple-choice format. Our results demonstrate that the presence of social identity cues does not merely shift predictions; it affects the reliability of confidence signals, posing a significant risk to equitable care and safe deployment in confidence-based clinical workflows.

cs.CL↗

arXiv2026

FewMMBench: A Benchmark for Multimodal Few-Shot Learning

As multimodal large language models (MLLMs) advance in handling interleaved image-text data, assessing their few-shot learning capabilities remains an open challenge. In this paper, we introduce FewMMBench, a comprehensive benchmark designed to evaluate MLLMs under few-shot conditions, with a focus on In-Context Learning (ICL) and Chain-of-Thought (CoT) prompting. Covering a diverse suite of multimodal understanding tasks, from attribute recognition to temporal reasoning, FewMMBench enables systematic analysis across task types, model families, and prompting strategies. We evaluate 26 open-weight MLLMs from six model families across zero-shot, few-shot, and CoT-augmented few-shot settings. Our findings reveal that instruction-tuned models exhibit strong zero-shot performance but benefit minimally, or even regress, with additional demonstrations or CoT reasoning. Retrieval-based demonstrations and increased context size also yield limited gains. These results highlight FewMMBench as a rigorous testbed for diagnosing and advancing few-shot capabilities in multimodal LLMs. The data is available at: https://huggingface.co/datasets/mustafaa/FewMMBench

cs.CL↗

arXiv2025

MedPath: Multi-Domain Cross-Vocabulary Hierarchical Paths for Biomedical Entity Linking

Progress in biomedical Named Entity Recognition (NER) and Entity Linking (EL) is currently hindered by a fragmented data landscape, a lack of resources for building explainable models, and the limitations of semantically-blind evaluation metrics. To address these challenges, we present MedPath, a large-scale and multi-domain biomedical EL dataset that builds upon nine existing expert-annotated EL datasets. In MedPath, all entities are 1) normalized using the latest version of the Unified Medical Language System (UMLS), 2) augmented with mappings to 62 other biomedical vocabularies and, crucially, 3) enriched with full ontological paths -- i.e., from general to specific -- in up to 11 biomedical vocabularies. MedPath directly enables new research frontiers in biomedical NLP, facilitating training and evaluation of semantic-rich and interpretable EL systems, and the development of the next generation of interoperable and explainable clinical NLP models.

cs.CL↗

arXiv2025

Detection of Adverse Drug Events in Dutch clinical free text documents using Transformer Models: benchmark study

In this study, we establish a benchmark for adverse drug event (ADE) detection in Dutch clinical free-text documents using several transformer models, clinical scenarios, and fit-for-purpose performance measures. We trained a Bidirectional Long Short-Term Memory (Bi-LSTM) model and four transformer-based Dutch and/or multilingual encoder models (BERTje, RobBERT, MedRoBERTa(.)nl, and NuNER) for the tasks of named entity recognition (NER) and relation classification (RC) using 102 richly annotated Dutch ICU clinical progress notes. Anonymized free-text clinical progress notes of patients admitted to the intensive care unit (ICU) of one academic hospital and discharge letters of patients admitted to Internal Medicine wards of two non-academic hospitals were reused. We evaluated our ADE RC models internally using the gold standard (two-step task) and predicted entities (end-to-end task). In addition, all models were externally validated for detecting ADEs at the document level. We report both micro- and macro-averaged F1 scores, given the dataset imbalance in ADEs. Although differences for the ADE RC task between the models were small, MedRoBERTa(.)nl was the best performing model with a macro-averaged F1 score of 0.63 using the gold standard and 0.62 using predicted entities. The MedRoBERTa(.)nl models also performed the best in our external validation and achieved a recall of between 0.67 to 0.74 using predicted entities, meaning between 67 to 74% of discharge letters with ADEs were detected. Our benchmark study presents a robust and clinically meaningful approach for evaluating language models for ADE detection in clinical free-text documents. Our study highlights the need to use appropriate performance measures fit for the task of ADE detection in clinical free-text documents and envisioned future clinical use.

cs.CL↗

arXiv2025

DistillNote: Toward a Functional Evaluation Framework of LLM-Generated Clinical Note Summaries

Large language models (LLMs) are increasingly used to generate summaries from clinical notes. However, their ability to preserve essential diagnostic information remains underexplored, which could lead to serious risks for patient care. This study introduces DistillNote, an evaluation framework for LLM summaries that targets their functional utility by applying the generated summary downstream in a complex clinical prediction task, explicitly quantifying how much prediction signal is retained. We generated over 192,000 LLM summaries from MIMIC-IV clinical notes with increasing compression rates: standard, section-wise, and distilled section-wise. Heart failure diagnosis was chosen as the prediction task, as it requires integrating a wide range of clinical signals. LLMs were fine-tuned on both the original notes and their summaries, and their diagnostic performance was compared using the AUROC metric. We contrasted DistillNote's results with evaluations from LLM-as-judge and clinicians, assessing consistency across different evaluation methods. Summaries generated by LLMs maintained a strong level of heart failure diagnostic signal despite substantial compression. Models trained on the most condensed summaries (about 20 times smaller) achieved an AUROC of 0.92, compared to 0.94 with the original note baseline (97 percent retention). Functional evaluation provided a new lens for medical summary assessment, emphasizing clinical utility as a key dimension of quality. DistillNote introduces a new scalable, task-based method for assessing the functional utility of LLM-generated clinical summaries. Our results detail compression-to-performance tradeoffs from LLM clinical summarization for the first time. The framework is designed to be adaptable to other prediction tasks and clinical domains, aiding data-driven decisions about deploying LLM summarizers in real-world healthcare settings.

cs.CL↗

arXiv2025

DeVisE: Behavioral Testing of Medical Large Language Models

Large language models (LLMs) are increasingly applied in clinical decision support, yet current evaluations rarely reveal whether their outputs reflect genuine medical reasoning or superficial correlations. We introduce DeVisE (Demographics and Vital signs Evaluation), a behavioral testing framework that probes fine-grained clinical understanding through controlled counterfactuals. Using intensive care unit (ICU) discharge notes from MIMIC-IV, we construct both raw (real-world) and template-based (synthetic) variants with single-variable perturbations in demographic (age, gender, ethnicity) and vital sign attributes. We evaluate eight LLMs, spanning general-purpose and medical variants, under zero-shot setting. Model behavior is analyzed through (1) input-level sensitivity, capturing how counterfactuals alter perplexity, and (2) downstream reasoning, measuring their effect on predicted ICU length-of-stay and mortality. Overall, our results show that standard task metrics obscure clinically relevant differences in model behavior, with models differing substantially in how consistently and proportionally they adjust predictions to counterfactual perturbations.

cs.CL↗

arXiv2025

Mind the Gap: Benchmarking LLM Uncertainty and Calibration with Specialty-Aware Clinical QA and Reasoning-Based Behavioural Features

Reliable uncertainty quantification (UQ) is essential when employing large language models (LLMs) in high-risk domains such as clinical question answering (QA). In this work, we evaluate uncertainty estimation methods for clinical QA focusing, for the first time, on eleven clinical specialties and six question types, and across ten open-source LLMs (general-purpose, biomedical, and reasoning models), alongside representative proprietary models. We analyze score-based UQ methods, present a case study introducing a novel lightweight method based on behavioral features derived from reasoning-oriented models, and examine conformal prediction as a complementary set-based approach. Our findings reveal that uncertainty reliability is not a monolithic property, but one that depends on clinical specialty and question type due to shifts in calibration and discrimination. Our results highlight the need to select or ensemble models based on their distinct, complementary strengths and clinical use.

cs.CL↗

arXiv2025

Interpretable phenotyping of Heart Failure patients with Dutch discharge letters

Objective: Heart failure (HF) patients present with diverse phenotypes affecting treatment and prognosis. This study evaluates models for phenotyping HF patients based on left ventricular ejection fraction (LVEF) classes, using structured and unstructured data, assessing performance and interpretability. Materials and Methods: The study analyzes all HF hospitalizations at both Amsterdam UMC hospitals (AMC and VUmc) from 2015 to 2023 (33,105 hospitalizations, 16,334 patients). Data from AMC were used for model training, and from VUmc for external validation. The dataset was unlabelled and included tabular clinical measurements and discharge letters. Silver labels for LVEF classes were generated by combining diagnosis codes, echocardiography results, and textual mentions. Gold labels were manually annotated for 300 patients for testing. Multiple Transformer-based (black-box) and Aug-Linear (white-box) models were trained and compared with baselines on structured and unstructured data. To evaluate interpretability, two clinicians annotated 20 discharge letters by highlighting information they considered relevant for LVEF classification. These were compared to SHAP and LIME explanations from black-box models and the inherent explanations of Aug-Linear models. Results: BERT-based and Aug-Linear models, using discharge letters alone, achieved the highest classification results (AUC=0.84 for BERT, 0.81 for Aug-Linear on external validation), outperforming baselines. Aug-Linear explanations aligned more closely with clinicians' explanations than post-hoc explanations on black-box models. Conclusions: Discharge letters emerged as the most informative source for phenotyping HF patients. Aug-Linear models matched black-box performance while providing clinician-aligned interpretability, supporting their use in transparent clinical decision-making.

cs.CL↗

arXiv2025

What Does Neuro Mean to Cardio? Investigating the Role of Clinical Specialty Data in Medical LLMs

In this paper, we introduce S-MedQA, an English medical question-answering (QA) dataset designed for benchmarking large language models (LLMs) in fine-grained clinical specialties. S-MedQA consists of over 24k examples, covering 15 medical specialties, with QA pairs that can have multiple specialty annotations, such as when a question is cross-disciplinary. The dataset is constructed using both machine and expert verification to maximize data availability and reliability. We use S-MedQA to investigate the role of clinical specialties in the knowledge-intensive scenario of medical QA. Our results show that training on data from a clinical specialty does not necessarily lead to the best performance on that specialty. Additionally, regardless of the specialty the LLM was fine-tuned on, token probabilities of clinically relevant terms consistently increase across all specialties. Based on these findings, we hypothesize that improvement gains, at least in our settings, are derived primarily from domain shifting (e.g., general to medical) rather than from injecting specialty-specific knowledge. This suggests a need to rethink the role of fine-tuning data in the medical domain. To encourage further advancements in the clinical NLP field, we release S-MedQA along with all the code required to reproduce our experiments for the research community.

cs.CL↗

arXiv2024

AnyMatch -- Efficient Zero-Shot Entity Matching with a Small Language Model

Entity matching (EM) is the problem of determining whether two records refer to same real-world entity, which is crucial in data integration, e.g., for product catalogs or address databases. A major drawback of many EM approaches is their dependence on labelled examples. We thus focus on the challenging setting of zero-shot entity matching where no labelled examples are available for an unseen target dataset. Recently, large language models (LLMs) have shown promising results for zero-shot EM, but their low throughput and high deployment cost limit their applicability and scalability. We revisit the zero-shot EM problem with AnyMatch, a small language model fine-tuned in a transfer learning setup. We propose several novel data selection techniques to generate fine-tuning data for our model, e.g., by selecting difficult pairs to match via an AutoML filter, by generating additional attribute-level examples, and by controlling label imbalance in the data. We conduct an extensive evaluation of the prediction quality and deployment cost of our model, in a comparison to thirteen baselines on nine benchmark datasets. We find that AnyMatch provides competitive prediction quality despite its small parameter size: it achieves the second-highest F1 score overall, and outperforms several other approaches that employ models with hundreds of billions of parameters. Furthermore, our approach exhibits major cost benefits: the average prediction quality of AnyMatch is within 4.4% of the state-of-the-art method MatchGPT with the proprietary trillion-parameter model GPT-4, yet AnyMatch requires four orders of magnitude less parameters and incurs a 3,899 times lower inference cost (in dollars per 1,000 tokens).

cs.CL↗

arXiv2024

Evaluating Linguistic Capabilities of Multimodal LLMs in the Lens of Few-Shot Learning

The linguistic capabilities of Multimodal Large Language Models (MLLMs) are critical for their effective application across diverse tasks. This study aims to evaluate the performance of MLLMs on the VALSE benchmark, focusing on the efficacy of few-shot In-Context Learning (ICL), and Chain-of-Thought (CoT) prompting. We conducted a comprehensive assessment of state-of-the-art MLLMs, varying in model size and pretraining datasets. The experimental results reveal that ICL and CoT prompting significantly boost model performance, particularly in tasks requiring complex reasoning and contextual understanding. Models pretrained on captioning datasets show superior zero-shot performance, while those trained on interleaved image-text data benefit from few-shot learning. Our findings provide valuable insights into optimizing MLLMs for better grounding of language in visual contexts, highlighting the importance of the composition of pretraining data and the potential of few-shot learning strategies to improve the reasoning abilities of MLLMs.

cs.CL↗

arXiv2023

LLM aided semi-supervision for Extractive Dialog Summarization

Generating high-quality summaries for chat dialogs often requires large labeled datasets. We propose a method to efficiently use unlabeled data for extractive summarization of customer-agent dialogs. In our method, we frame summarization as a question-answering problem and use state-of-the-art large language models (LLMs) to generate pseudo-labels for a dialog. We then use these pseudo-labels to fine-tune a chat summarization model, effectively transferring knowledge from the large LLM into a smaller specialized model. We demonstrate our method on the \tweetsumm dataset, and show that using 10% of the original labelled data set we can achieve 65.9/57.0/61.0 ROUGE-1/-2/-L, whereas the current state-of-the-art trained on the entire training data set obtains 65.16/55.81/64.37 ROUGE-1/-2/-L. In other words, in the worst case (i.e., ROUGE-L) we still effectively retain 94.7% of the performance while using only 10% of the data.

cs.CL↗

arXiv2023

ViLMA: A Zero-Shot Benchmark for Linguistic and Temporal Grounding in Video-Language Models

With the ever-increasing popularity of pretrained Video-Language Models (VidLMs), there is a pressing need to develop robust evaluation methodologies that delve deeper into their visio-linguistic capabilities. To address this challenge, we present ViLMA (Video Language Model Assessment), a task-agnostic benchmark that places the assessment of fine-grained capabilities of these models on a firm footing. Task-based evaluations, while valuable, fail to capture the complexities and specific temporal aspects of moving images that VidLMs need to process. Through carefully curated counterfactuals, ViLMA offers a controlled evaluation suite that sheds light on the true potential of these models, as well as their performance gaps compared to human-level understanding. ViLMA also includes proficiency tests, which assess basic capabilities deemed essential to solving the main counterfactual tests. We show that current VidLMs' grounding abilities are no better than those of vision-language models which use static images. This is especially striking once the performance on proficiency tests is factored in. Our benchmark serves as a catalyst for future research on VidLMs, helping to highlight areas that still need to be explored.

cs.CL↗

arXiv2023

Fixing confirmation bias in feature attribution methods via semantic match

Feature attribution methods have become a staple method to disentangle the complex behavior of black box models. Despite their success, some scholars have argued that such methods suffer from a serious flaw: they do not allow a reliable interpretation in terms of human concepts. Simply put, visualizing an array of feature contributions is not enough for humans to conclude something about a model's internal representations, and confirmation bias can trick users into false beliefs about model behavior. We argue that a structured approach is required to test whether our hypotheses on the model are confirmed by the feature attributions. This is what we call the "semantic match" between human concepts and (sub-symbolic) explanations. Building on the conceptual framework put forward in Cin\`a et al. [2023], we propose a structured approach to evaluate semantic match in practice. We showcase the procedure in a suite of experiments spanning tabular and image data, and show how the assessment of semantic match can give insight into both desirable (e.g., focusing on an object relevant for prediction) and undesirable model behaviors (e.g., focusing on a spurious correlation). We couple our experimental results with an analysis on the metrics to measure semantic match, and argue that this approach constitutes the first step towards resolving the issue of confirmation bias in XAI.

cs.LG↗