We present ReXVQA, the largest and most comprehensive benchmark for visual question answering (VQA) in chest radiology, comprising approximately 696,000 questions paired with 160,000 chest X-rays studies across training, validation, and test sets. Unlike prior efforts that rely heavily on template based queries, ReXVQA introduces a diverse and clinically authentic task suite reflecting five core radiological reasoning skills: presence assessment, location analysis, negation detection, differential diagnosis, and geometric reasoning. We evaluate eight state-of-the-art multimodal large language models, including MedGemma-4B-it, Qwen2.5-VL, Janus-Pro-7B, and Eagle2-9B. The best-performing model (MedGemma) achieves 83.24% overall accuracy. To bridge the gap between AI performance and clinical expertise, we conducted a comprehensive human reader study involving 3 radiology residents on 200 randomly sampled cases. Our evaluation demonstrates that MedGemma achieved superior performance (83.84% accuracy) compared to human readers (best radiology resident: 77.27%), representing a significant milestone where AI performance exceeds expert human evaluation on chest X-ray interpretation. The reader study reveals distinct performance patterns between AI models and human experts, with strong inter-reader agreement among radiologists while showing more variable agreement patterns between human readers and AI models. ReXVQA establishes a new standard for evaluating generalist radiological AI systems, offering public leaderboards, fine-grained evaluation splits, structured explanations, and category-level breakdowns. This benchmark lays the foundation for next-generation AI systems capable of mimicking expert-level clinical reasoning beyond narrow pathology classification. Our dataset will be open-sourced at https://huggingface.co/datasets/rajpurkarlab/ReXVQA
Assertion status detection is a critical yet often overlooked component of clinical NLP, essential for accurately attributing extracted medical facts. Past studies have narrowly focused on negation detection, leading to underperforming commercial solutions such as AWS Medical Comprehend, Azure AI Text Analytics, and GPT-4o due to their limited domain adaptation. To address this gap, we developed state-of-the-art assertion detection models, including fine-tuned LLMs, transformer-based classifiers, few-shot classifiers, and deep learning (DL) approaches. We evaluated these models against cloud-based commercial API solutions, the legacy rule-based NegEx approach, and GPT-4o. Our fine-tuned LLM achieves the highest overall accuracy (0.962), outperforming GPT-4o (0.901) and commercial APIs by a notable margin, particularly excelling in Present (+4.2%), Absent (+8.4%), and Hypothetical (+23.4%) assertions. Our DL-based models surpass commercial solutions in Conditional (+5.3%) and Associated-with-Someone-Else (+10.1%) categories, while the few-shot classifier offers a lightweight yet highly competitive alternative (0.929), making it ideal for resource-constrained environments. Integrated within Spark NLP, our models consistently outperform black-box commercial solutions while enabling scalable inference and seamless integration with medical NER, Relation Extraction, and Terminology Resolution. These results reinforce the importance of domain-adapted, transparent, and customizable clinical NLP solutions over general-purpose LLMs and proprietary APIs.
In this work, we measure the impact of affixal negation on modern English large language models (LLMs). In affixal negation, the negated meaning is expressed through a negative morpheme, which is potentially challenging for LLMs as their tokenizers are often not morphologically plausible. We conduct extensive experiments using LLMs with different subword tokenization methods, which lead to several insights on the interaction between tokenization performance and negation sensitivity. Despite some interesting mismatches between tokenization accuracy and negation detection performance, we show that models can, on the whole, reliably recognize the meaning of affixal negation.
Clinical trials (CTs) often fail due to inadequate patient recruitment. This paper tackles the challenges of CT retrieval by presenting an approach that addresses the patient-to-trials paradigm. Our approach involves two key components in a pipeline-based model: (i) a data enrichment technique for enhancing both queries and documents during the first retrieval stage, and (ii) a novel re-ranking schema that uses a Transformer network in a setup adapted to this task by leveraging the structure of the CT documents. We use named entity recognition and negation detection in both patient description and the eligibility section of CTs. We further classify patient descriptions and CT eligibility criteria into current, past, and family medical conditions. This extracted information is used to boost the importance of disease and drug mentions in both query and index for lexical retrieval. Furthermore, we propose a two-step training schema for the Transformer network used to re-rank the results from the lexical retrieval. The first step focuses on matching patient information with the descriptive sections of trials, while the second step aims to determine eligibility by matching patient information with the criteria section. Our findings indicate that the inclusion criteria section of the CT has a great influence on the relevance score in lexical models, and that the enrichment techniques for queries and documents improve the retrieval of relevant trials. The re-ranking strategy, based on our training schema, consistently enhances CT retrieval and shows improved performance by 15\% in terms of precision at retrieving eligible trials. The results of our experiments suggest the benefit of making use of extracted entities. Moreover, our proposed re-ranking schema shows promising effectiveness compared to larger neural models, even with limited training data.
Negation is a fundamental aspect of natural language, playing a critical role in communication and comprehension. Our study assesses the negation detection performance of Generative Pre-trained Transformer (GPT) models, specifically GPT-2, GPT-3, GPT-3.5, and GPT-4. We focus on the identification of negation in natural language using a zero-shot prediction approach applied to our custom xNot360 dataset. Our approach examines sentence pairs labeled to indicate whether the second sentence negates the first. Our findings expose a considerable performance disparity among the GPT models, with GPT-4 surpassing its counterparts and GPT-3.5 displaying a marked performance reduction. The overall proficiency of the GPT models in negation detection remains relatively modest, indicating that this task pushes the boundaries of their natural language understanding capabilities. We not only highlight the constraints of GPT models in handling negation but also emphasize the importance of logical reliability in high-stakes domains such as healthcare, science, and law.
This study aims to demonstrate the methods for detecting negations in a sentence by uniquely evaluating the lexical structure of the text via word-sense disambiguation. The proposed framework examines all the unique features in the various expressions within a text to resolve the contextual usage of all tokens and decipher the effect of negation on sentiment analysis. The application of popular expression detectors skips this important step, thereby neglecting the root words caught in the web of negation and making text classification difficult for machine learning and sentiment analysis. This study adopts the Natural Language Processing (NLP) approach to discover and antonimize words that were negated for better accuracy in text classification using a knowledge base provided by an NLP library called WordHoard. Early results show that our initial analysis improved on traditional sentiment analysis, which sometimes neglects negations or assigns an inverse polarity score. The SentiWordNet analyzer was improved by 35%, the Vader analyzer by 20% and the TextBlob by 6%.
As structured data are often insufficient, labels need to be extracted from free text in electronic health records when developing models for clinical information retrieval and decision support systems. One of the most important contextual properties in clinical text is negation, which indicates the absence of findings. We aimed to improve large scale extraction of labels by comparing three methods for negation detection in Dutch clinical notes. We used the Erasmus Medical Center Dutch Clinical Corpus to compare a rule-based method based on ContextD, a biLSTM model using MedCAT and (finetuned) RoBERTa-based models. We found that both the biLSTM and RoBERTa models consistently outperform the rule-based model in terms of F1 score, precision and recall. In addition, we systematically categorized the classification errors for each model, which can be used to further improve model performance in particular applications. Combining the three models naively was not beneficial in terms of performance. We conclude that the biLSTM and RoBERTa-based models in particular are highly accurate accurate in detecting clinical negations, but that ultimately all three approaches can be viable depending on the use case at hand.
Negation is a common linguistic feature that is crucial in many language understanding tasks, yet it remains a hard problem due to diversity in its expression in different types of text. Recent work has shown that state-of-the-art NLP models underperform on samples containing negation in various tasks, and that negation detection models do not transfer well across domains. We propose a new negation-focused pre-training strategy, involving targeted data augmentation and negation masking, to better incorporate negation information into language models. Extensive experiments on common benchmarks show that our proposed approach improves negation detection performance and generalizability over the strong baseline NegBERT (Khandewal and Sawant, 2020).
Analyzing radiology reports is a time-consuming and error-prone task, which raises the need for an efficient automated radiology report analysis system to alleviate the workloads of radiologists and encourage precise diagnosis. In this work, we present RadText, an open-source radiology text analysis system developed by Python. RadText offers an easy-to-use text analysis pipeline, including de-identification, section segmentation, sentence split and word tokenization, named entity recognition, parsing, and negation detection. RadText features a flexible modular design, provides a hybrid text processing schema, and supports raw text processing and local processing, which enables better usability and improved data privacy. RadText adopts BioC as the unified interface, and also standardizes the input / output into a structured representation compatible with Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM). This allows for a more systematic approach to observational research across multiple, disparate data sources. We evaluated RadText on the MIMIC-CXR dataset, with five new disease labels we annotated for this work. RadText demonstrates highly accurate classification performances, with an average precision of, a recall of 0.94, and an F-1 score of 0.92. We have made our code, documentation, examples, and the test set available at https://github.com/bionlplab/radtext .
Adverse Drug Event (ADE) extraction models can rapidly examine large collections of social media texts, detecting mentions of drug-related adverse reactions and trigger medical investigations. However, despite the recent advances in NLP, it is currently unknown if such models are robust in face of negation, which is pervasive across language varieties. In this paper we evaluate three state-of-the-art systems, showing their fragility against negation, and then we introduce two possible strategies to increase the robustness of these models: a pipeline approach, relying on a specific component for negation detection; an augmentation of an ADE extraction dataset to artificially create negated samples and further train the models. We show that both strategies bring significant increases in performance, lowering the number of spurious entities predicted by the models. Our dataset and code will be publicly released to encourage research on the topic.
Neural network-based methods are the state of the art in negation scope resolution. However, they often use the unrealistic assumption that cue information is completely accurate. Even if this assumption holds, there remains a dependency on engineered features from state-of-the-art machine learning methods. The current study adopted a two-step negation resolving apporach to assess whether a Bidirectional Long Short-Term Memory-based method can be used for cue detection as well, and how inaccurate cue predictions would affect the scope resolution performance. Results suggest that this method is not suitable for negation detection. Scope resolution performance is most robust against inaccurate information for models with a recurrent layer only, compared to extensions with a Conditional Random Fields layer or a post-processing algorithm. We advocate for more research into the application of deep learning on negation detection and the effect of imperfect information on scope resolution.
Genome-wide association (GWA) constitutes a prominent portion of studies which have been conducted on personalized medicine and pharmacogenomics. Recently, very few methods have been developed for extracting mutation-diseases associations. However, there is no available method for extracting the association of SNP-phenotype from text which considers degree of confidence in associations. In this study, first a relation extraction method relying on linguistic-based negation detection and neutral candidates is proposed. The experiments show that negation cues and scope as well as detecting neutral candidates can be employed for implementing a superior relation extraction method which outperforms the kernel-based counterparts due to a uniform innate polarity of sentences and small number of complex sentences in the corpus. Moreover, a modality based approach is proposed to estimate the confidence level of the extracted association which can be used to assess the reliability of the reported association. Keywords: SNP, Phenotype, Biomedical Relation Extraction, Negation Detection.
Developing predictive modeling in medicine requires additional features from unstructured clinical texts. In Russia, there are no instruments for natural language processing to cope with problems of medical records. This paper is devoted to a module of negation detection. The corpus-free machine learning method is based on gradient boosting classifier is used to detect whether a disease is denied, not mentioned or presented in the text. The detector classifies negations for five diseases and shows average F-score from 0.81 to 0.93. The benefits of negation detection have been demonstrated by predicting the presence of surgery for patients with the acute coronary syndrome.
In the era of clinical information explosion, a good strategy for clinical text summarization is helpful to improve the clinical workflow. The ideal summarization strategy can preserve important information in the informative but less organized, ill-structured clinical narrative texts. Instead of using pure statistical learning approaches, which are difficult to interpret and explain, we utilized knowledge of computational linguistics with human experts-curated biomedical knowledge base to achieve the interpretable and meaningful clinical text summarization. Our research objective is to use the biomedical ontology with semantic information, and take the advantage from the language hierarchical structure, the constituency tree, in order to identify the correct clinical concepts and the corresponding negation information, which is critical for summarizing clinical concepts from narrative text. We achieved the clinically acceptable performance for both negation detection and concept identification, and the clinical concepts with common negated patterns can be identified and negated by the proposed method.
Speculation is a naturally occurring phenomena in textual data, forming an integral component of many systems, especially in the biomedical information retrieval domain. Previous work addressing cue detection and scope resolution (the two subtasks of speculation detection) have ranged from rule-based systems to deep learning-based approaches. In this paper, we apply three popular transformer-based architectures, BERT, XLNet and RoBERTa to this task, on two publicly available datasets, BioScope Corpus and SFU Review Corpus, reporting substantial improvements over previously reported results (by at least 0.29 F1 points on cue detection and 4.27 F1 points on scope resolution). We also experiment with joint training of the model on multiple datasets, which outperforms the single dataset training approach by a good margin. We observe that XLNet consistently outperforms BERT and RoBERTa, contrary to results on other benchmark datasets. To confirm this observation, we apply XLNet and RoBERTa to negation detection and scope resolution, reporting state-of-the-art results on negation scope resolution for the BioScope Corpus (increase of 3.16 F1 points on the BioScope Full Papers, 0.06 F1 points on the BioScope Abstracts) and the SFU Review Corpus (increase of 0.3 F1 points).
Negation is an important characteristic of language, and a major component of information extraction from text. This subtask is of considerable importance to the biomedical domain. Over the years, multiple approaches have been explored to address this problem: Rule-based systems, Machine Learning classifiers, Conditional Random Field Models, CNNs and more recently BiLSTMs. In this paper, we look at applying Transfer Learning to this problem. First, we extensively review previous literature addressing Negation Detection and Scope Resolution across the 3 datasets that have gained popularity over the years: the BioScope Corpus, the Sherlock dataset, and the SFU Review Corpus. We then explore the decision choices involved with using BERT, a popular transfer learning model, for this task, and report state-of-the-art results for scope resolution across all 3 datasets. Our model, referred to as NegBERT, achieves a token level F1 score on scope resolution of 92.36 on the Sherlock dataset, 95.68 on the BioScope Abstracts subcorpus, 91.24 on the BioScope Full Papers subcorpus, 90.95 on the SFU Review Corpus, outperforming the previous state-of-the-art systems by a significant margin. We also analyze the model's generalizability to datasets on which it is not trained.
Recognizing the entailment relation showed that its influence to extract the semantic inferences in wide-ranging natural language processing domains (text summarization, question answering, etc.) and enhanced the results of their output. For Arabic language, few attempts concerns with Arabic entailment problem. This paper aims to increase the entailment accuracy for Arabic texts by resolving negation of the text-hypothesis pair and determining the polarity of the text-hypothesis pair whether it is Positive, Negative or Neutral. It is noticed that the absence of negation detection feature gives inaccurate results when detecting the entailment relation since the negation revers the truth. The negation words are considered stop words and removed from the text-hypothesis pair which may lead wrong entailment decision. Another case not solved previously, it is impossible that the positive text entails negative text and vice versa. In this paper, in order to classify the text-hypothesis pair polarity, a sentiment analysis tool is used. We show that analyzing the polarity of the text-hypothesis pair increases the entailment accuracy. to evaluate our approach we used a dataset for Arabic textual entailment (ArbTEDS) consisted of 618 text-hypothesis pairs and showed that the Arabic entailment accuracy is increased by resolving negation for entailment relation and analyzing the polarity of the text-hypothesis pair.
Negative medical findings are prevalent in clinical reports, yet discriminating them from positive findings remains a challenging task for information extraction. Most of the existing systems treat this task as a pipeline of two separate tasks, i.e., named entity recognition (NER) and rule-based negation detection. We consider this as a multi-task problem and present a novel end-to-end neural model to jointly extract entities and negations. We extend a standard hierarchical encoder-decoder NER model and first adopt a shared encoder followed by separate decoders for the two tasks. This architecture performs considerably better than the previous rule-based and machine learning-based systems. To overcome the problem of increased parameter size especially for low-resource settings, we propose the Conditional Softmax Shared Decoder architecture which achieves state-of-art results for NER and negation detection on the 2010 i2b2/VA challenge dataset and a proprietary de-identified clinical dataset.
Clinical text provides essential information to estimate the acuity of a patient during hospital stays in addition to structured clinical data. In this study, we explore how clinical text can complement a clinical predictive learning task. We leverage an internal medical natural language processing service to perform named entity extraction and negation detection on clinical notes and compose selected entities into a new text corpus to train document representations. We then propose a multimodal neural network to jointly train time series signals and unstructured clinical text representations to predict the in-hospital mortality risk for ICU patients. Our model outperforms the benchmark by 2% AUC.
Motivated by the need to automate medical information extraction from free-text radiological reports, we present a bi-directional long short-term memory (BiLSTM) neural network architecture for modelling radiological language. The model has been used to address two NLP tasks: medical named-entity recognition (NER) and negation detection. We investigate whether learning several types of word embeddings improves BiLSTM's performance on those tasks. Using a large dataset of chest x-ray reports, we compare the proposed model to a baseline dictionary-based NER system and a negation detection system that leverages the hand-crafted rules of the NegEx algorithm and the grammatical relations obtained from the Stanford Dependency Parser. Compared to these more traditional rule-based systems, we argue that BiLSTM offers a strong alternative for both our tasks.