The rapid expansion of biomedical publications creates challenges for organizing knowledge and detecting emerging trends, underscoring the need for scalable and interpretable methods. Common clustering and topic modeling approaches such as K-means or LDA remain sensitive to initialization and prone to local optima, limiting reproducibility and evaluation. We propose a reformulation of a convex-optimization-based clustering algorithm that produces stable, fine-grained topics by selecting exemplars from the data and guaranteeing a global optimum. Applied to ~12,000 PubMed articles on aging and longevity, our method uncovers topics validated by medical experts. It yields interpretable topics spanning from molecular mechanisms to dietary supplements, physical activity, and gut microbiota. The method performs favorably, and most importantly, its reproducibility and interpretability distinguish it from common clustering approaches, including K-means, LDA, and BERTopic. This work provides a basis for developing scalable, web-accessible tools for knowledge discovery.

Evaluating large language models (LLMs) in the biomedical domain requires benchmarks that can distinguish genuine reasoning from pattern matching and that remain discriminative as model capabilities improve. Existing biomedical question answering (QA) benchmarks are limited in this respect: multiple-choice formats allow models to succeed by answer elimination rather than inference, and widely circulated exam-style datasets are subject to performance saturation and training data contamination. Multi-hop reasoning. i.e., the ability to integrate information across multiple sources to derive an answer, is central to clinically meaningful tasks such as diagnosis support, literature-based discovery, and hypothesis generation, yet remains underrepresented in current biomedical QA benchmarks. We present MedHopQA, a disease-centered multi-hop reasoning benchmark of 1,000 expert-curated question-answer pairs introduced as a shared task at BioCreative IX. Each question requires synthesis of information across two distinct Wikipedia articles, and answers are provided in open-ended free-text format rather than as multiple-choice selections. Gold annotations are augmented with ontology-grounded synonym sets (MONDO, NCBI Gene, NCBI Taxonomy) to support both lexical and concept-level evaluation. The dataset was constructed through a multi-stage human-AI pipeline combining structured human annotation, triage, iterative verification, and LLM-as-a-judge validation. To reduce leaderboard gaming and contamination risk, the 1,000 scored questions are embedded within a publicly downloadable set of 10,000 questions, with answers withheld, on a CodaBench leaderboard. Evaluation of four frontier LLMs under a zero-shot setting (GPT-5.1, Gemini 2.5 Pro, Claude Sonnet 4.5, and GPT-4o) reveals performance variation across answer types, with overall accuracy ranging from 66.3% to 83.4%. Performance is strongest on chemical and anatomical questions and most variable on disease and gene/protein categories, where fine-grained semantic discrimination is required. MedHopQA provides both a benchmark and a reusable framework for constructing future biomedical QA datasets that prioritize compositional reasoning, saturation resistance, and contamination mitigation as design constraints.

LitSense 2.0 (https://www.ncbi.nlm.nih.gov/research/litsense2/) is an advanced biomedical search system enhanced with dense vector semantic retrieval, designed for accessing literature on sentence and paragraph levels. It provides unified access to 38 million PubMed abstracts and 6.6 million full-length articles in the PubMed Central (PMC) Open Access subset, encompassing 1.4 billion sentences and ∼300 million paragraphs, and is updated weekly. Compared to PubMed and PMC, the primary platforms for biomedical information search, LitSense offers cross-platform functionality by searching seamlessly across both PubMed and PMC and returning relevant results at a more granular level. Building on the success of the original LitSense launched in 2018, LitSense 2.0 introduces two major enhancements. The first is the addition of paragraph-level search: users can now choose to search either against sentences or against paragraphs. The second is improved retrieval accuracy via a state-of-the-art biomedical text encoder, ensuring more reliable identification of relevant results across the entire biomedical literature.

Clinical calculators play a vital role in healthcare, but their utilization is often hindered by usability and dissemination challenges. We introduce AgentMD, a novel language agent capable of curating and applying clinical calculators across various clinical contexts. As a tool builder, AgentMD first uses PubMed to curate a diverse set of 2,164 executable clinical calculators with over 85% accuracy for quality checks and over 90% pass rate for unit tests. As a tool user, AgentMD autonomously selects and applies the relevant clinical calculators. Our evaluations show that AgentMD significantly outperforms GPT-4 for risk prediction (87.7% vs. 40.9% in accuracy). Results on 698 real-world emergency department notes confirm that AgentMD accurately computes medical risks at an individual level. Moreover, AgentMD can provide population-level insights for institutional risk management. Our study illustrates the capabilities of language agents to curate and utilize clinical calculators for both individual patient care and at-scale healthcare analytics.

SUMMARY:Over 55% of author names in PubMed are ambiguous: the same name is shared by different individual researchers. This poses significant challenges on precise literature retrieval for author name queries, a common behavior in biomedical literature search. In response, we present a comprehensive dataset of disambiguated authors. Specifically, we complement the automatic PubMed Computed Authors algorithm with the latest ORCID data for improved accuracy. As a result, the enhanced algorithm achieves high performance in author name disambiguation, and subsequently our dataset contains more than 21 million disambiguated authors for over 35 million PubMed articles and is incrementally updated on a weekly basis. More importantly, we make the dataset publicly available for the community such that it can be utilized in a wide variety of potential applications beyond assisting PubMed's author name queries. Finally, we propose a set of guidelines for best practices of authors pertaining to use of their names. AVAILABILITY AND IMPLEMENTATION:The PubMed Computed Authors dataset is publicly available for bulk download at: https://ftp.ncbi.nlm.nih.gov/pub/lu/ComputedAuthors/. Additionally, it is available for query through web API at: https://www.ncbi.nlm.nih.gov/research/bionlp/APIs/authors/.

The remarkable capabilities of Large Language Models (LLMs) make them increasingly compelling for adoption in real-world healthcare applications. However, the risks associated with using LLMs in medical applications have not been systematically characterized. We propose using five key principles for safe and trustworthy medical AI: Truthfulness, Resilience, Fairness, Robustness, and Privacy, along with ten specific aspects. Under this comprehensive framework, we introduce a novel MedGuard benchmark with 1,000 expert-verified questions. Our evaluation of 11 commonly used LLMs shows that the current language models, regardless of their safety alignment mechanisms, generally perform poorly on most of our benchmarks, particularly when compared to the high performance of human physicians. Despite recent reports indicate that advanced LLMs like ChatGPT can match or even exceed human performance in various medical tasks, this study underscores a significant safety gap, highlighting the crucial need for human oversight and the implementation of AI safety guardrails.

MOTIVATION:Information retrieval (IR) is essential in biomedical knowledge acquisition and clinical decision support. While recent progress has shown that language model encoders perform better semantic retrieval, training such models requires abundant query-article annotations that are difficult to obtain in biomedicine. As a result, most biomedical IR systems only conduct lexical matching. In response, we introduce MedCPT, a first-of-its-kind Contrastively Pre-trained Transformer model for zero-shot semantic IR in biomedicine. RESULTS:To train MedCPT, we collected an unprecedented scale of 255 million user click logs from PubMed. With such data, we use contrastive learning to train a pair of closely integrated retriever and re-ranker. Experimental results show that MedCPT sets new state-of-the-art performance on six biomedical IR tasks, outperforming various baselines including much larger models, such as GPT-3-sized cpt-text-XL. In addition, MedCPT also generates better biomedical article and sentence representations for semantic evaluations. As such, MedCPT can be readily applied to various real-world biomedical IR tasks. AVAILABILITY AND IMPLEMENTATION:The MedCPT code and model are available at https://github.com/ncbi/MedCPT.

OBJECTIVE:A significant number of recent articles in PubMed have full text available in PubMed Central®, and the availability of full texts has been consistently growing. However, it is not currently possible for a user to simultaneously query the contents of both databases and receive a single integrated search result. In this study, we investigate how to score full text articles given a multitoken query and how to combine those full text article scores with scores originating from abstracts and achieve an overall improved retrieval performance. MATERIALS AND METHODS:For scoring full text articles, we propose a method to combine information coming from different sections by converting the traditionally used BM25 scores into log odds ratio scores which can be treated uniformly. We further propose a method that successfully combines scores from two heterogenous retrieval sources - full text articles and abstract only articles - by balancing the contributions of their respective scores through a probabilistic transformation. We use PubMed click data that consists of queries sampled from PubMed user logs along with a subset of retrieved and clicked documents to train the probabilistic functions and to evaluate retrieval effectiveness. RESULTS AND CONCLUSIONS:Random ranking achieves 0.579 MAP score on our PubMed click data. BM25 ranking on PubMed abstracts improves the MAP by 10.6%. For full text documents, experiments confirm that BM25 section scores are of different value depending on the section type and are not directly comparable. Naïvely using the body text of articles along with abstract text degrades the overall quality of the search. The proposed log odds ratio scores normalize and combine the contributions of occurrences of query tokens in different sections. By including full text where available, we gain another 0.67%, or 7% relative improvement over abstract alone. We find an advantage in the more accurate estimate of the value of BM25 scores depending on the section from which they were produced. Taking the sum of top three section scores performs the best.

A significant percentage of COVID-19 survivors experience ongoing multisystemic symptoms that often affect daily living, a condition known as Long Covid or post-acute-sequelae of SARS-CoV-2 infection. However, identifying scientific articles relevant to Long Covid is challenging since there is no standardized or consensus terminology. We developed an iterative human-in-the-loop machine learning framework combining data programming with active learning into a robust ensemble model, demonstrating higher specificity and considerably higher sensitivity than other methods. Analysis of the Long Covid Collection shows that (1) most Long Covid articles do not refer to Long Covid by any name, (2) when the condition is named, the name used most frequently in the literature is Long Covid, and (3) Long Covid is associated with disorders in a wide variety of body systems. The Long Covid Collection is updated weekly and is searchable online at the LitCovid portal: https://www.ncbi.nlm.nih.gov/research/coronavirus/docsum?filters=e_condition.LongCovid.

To inform continuous and rigorous reflection about the description of human populations in genomics research, this study investigates the historical and contemporary use of the terms "ancestry," "ethnicity," "race," and other population labels in The American Journal of Human Genetics from 1949 to 2018. We characterize these terms' frequency of use and assess their odds of co-occurrence with a set of social and genetic topical terms. Throughout The Journal's 70-year history, "ancestry" and "ethnicity" have increased in use, appearing in 33% and 26% of articles in 2009-2018, while the use of "race" has decreased, occurring in 4% of articles in 2009-2018. Although its overall use has declined, the odds of "race" appearing in the presence of "ethnicity" has increased relative to the odds of occurring in its absence. Forms of population descriptors "Caucasian" and "Negro" have largely disappeared from The Journal (<1% of articles in 2009-2018). Conversely, the continental labels "African," "Asian," and "European" have increased in use and appear in 18%, 14%, and 42% of articles from 2009-2018, respectively. Decreasing uses of the terms "race," "Caucasian," and "Negro" are indicative of a transition away from the field's history of explicitly biological race science; at the same time, the increasing use of "ancestry," "ethnicity," and continental labels should serve to motivate ongoing reflection as the terminology used to describe genetic variation continues to evolve.

With the growing availability of full-text articles, integrating abstracts and full texts of documents into a unified representation is essential for comprehensive search of scientific literature. However, previous studies have shown that naïvely merging abstracts with full texts of articles does not consistently yield better performance. Balancing the contribution of query terms appearing in the abstract and in sections of different importance in full text articles remains a challenge both with traditional bag-of-words IR approaches and for neural retrieval methods. In this work we establish the connection between the BM25 score of a query term appearing in a section of a full text document and the probability of that document being clicked or identified as relevant. Probability is computed using Pool Adjacent Violators (PAV), an isotonic regression algorithm, providing a maximum likelihood estimate based on the observed data. Using this probabilistic transformation of BM25 scores we show an improved performance on the PubMed Click dataset developed and presented in this study, as well as the 2007 TREC Genomics collection.

OBJECTIVE In a biomedical literature search, the link between a query and a document is often not established, because they use different terms to refer to the same concept. Distributional word embeddings are frequently used for detecting related words by computing the cosine similarity between them. However, previous research has not established either the best embedding methods for detecting synonyms among related word pairs or how effective such methods may be. MATERIALS AND METHODS In this study, we first create the BioSearchSyn set, a manually annotated set of synonyms, to assess and compare 3 widely used word-embedding methods (word2vec, fastText, and GloVe) in their ability to detect synonyms among related pairs of words. We demonstrate the shortcomings of the cosine similarity score between word embeddings for this task: the same scores have very different meanings for the different methods. To address the problem, we propose utilizing pool adjacent violators (PAV), an isotonic regression algorithm, to transform a cosine similarity into a probability of 2 words being synonyms. RESULTS Experimental results using the BioSearchSyn set as a gold standard reveal which embedding methods have the best performance in identifying synonym pairs. The BioSearchSyn set also allows converting cosine similarity scores into probabilities, which provides a uniform interpretation of the synonymy score over different methods. CONCLUSIONS We introduced the BioSearchSyn corpus of 1000 term pairs, which allowed us to identify the best embedding method for detecting synonymy for biomedical search. Using the proposed method, we created PubTermVariants2.0: a large, automatically extracted set of synonym pairs that have augmented PubMed searches since the spring of 2019.

The need to organize a large collection in a manner that facilitates human comprehension is crucial given the ever-increasing volumes of information. In this work, we present PDC (probabilistic distributional clustering), a novel algorithm that, given a document collection, computes disjoint term sets representing topics in the collection. The algorithm relies on probabilities of word co-occurrences to partition the set of terms appearing in the collection of documents into disjoint groups of related terms. In this work, we also present an environment to visualize the computed topics in the term space and retrieve the most related PubMed articles for each group of terms. We illustrate the algorithm by applying it to PubMed documents on the topic of suicide. Suicide is a major public health problem identified as the tenth leading cause of death in the US. In this application, our goal is to provide a global view of the mental health literature pertaining to the subject of suicide, and through this, to help create a rich environment of multifaceted data to guide health care researchers in their endeavor to better understand the breadth, depth and scope of the problem. We demonstrate the usefulness of the proposed algorithm by providing a web portal that allows mental health researchers to peruse the suicide-related literature in PubMed.

Background Capturing sentence semantics plays a vital role in a range of text mining applications. Despite continuous efforts on the development of related datasets and models in the general domain, both datasets and models are limited in biomedical and clinical domains. The BioCreative/OHNLP2018 organizers have made the first attempt to annotate 1068 sentence pairs from clinical notes and have called for a community effort to tackle the Semantic Textual Similarity (BioCreative/OHNLP STS) challenge. Methods We developed models using traditional machine learning and deep learning approaches. For the post challenge, we focused on two models: the Random Forest and the Encoder Network. We applied sentence embeddings pre-trained on PubMed abstracts and MIMIC-III clinical notes and updated the Random Forest and the Encoder Network accordingly. Results The official results demonstrated our best submission was the ensemble of eight models. It achieved a Person correlation coefficient of 0.8328 – the highest performance among 13 submissions from 4 teams. For the post challenge, the performance of both Random Forest and the Encoder Network was improved; in particular, the correlation of the Encoder Network was improved by ~ 13%. During the challenge task, no end-to-end deep learning models had better performance than machine learning models that take manually-crafted features. In contrast, with the sentence embeddings pre-trained on biomedical corpora, the Encoder Network now achieves a correlation of ~ 0.84, which is higher than the original best model. The ensembled model taking the improved versions of the Random Forest and Encoder Network as inputs further increased performance to 0.8528. Conclusions Deep learning models with sentence embeddings pre-trained on biomedical corpora achieve the highest performance on the test set. Through error analysis, we find that end-to-end deep learning models and traditional machine learning models with manually-crafted features complement each other by finding different types of sentences. We suggest a combination of these models can better find similar sentences in practice.

Timely access to accurate scientific literature in the battle with the ongoing COVID-19 pandemic is critical. This unprecedented public health risk has motivated research towards understanding the disease in general, identifying drugs to treat the disease, developing potential vaccines, etc. This has given rise to a rapidly growing body of literature that doubles in number of publications every 20 days as of May 2020. Providing medical professionals with means to quickly analyze the literature and discover growing areas of knowledge is necessary for addressing their question and information needs. In this study we analyze the LitCovid collection, 13,369 COVID-19 related articles found in PubMed as of May 15th, 2020 with the purpose of examining the landscape of literature and presenting it in a format that facilitates information navigation and understanding. We do that by applying state-of-the-art named entity recognition, classification, clustering and other NLP techniques. By applying NER tools, we capture relevant bioentities (such as diseases, internal body organs, etc.) and assess the strength of their relationship with COVID-19 by the extent they are discussed in the corpus. We also collect a variety of symptoms and co-morbidities discussed in reference to COVID-19. Our clustering algorithm identifies topics represented by groups of related terms, and computes clusters corresponding to documents associated with the topic terms. Among the topics we observe several that persist through the duration of multiple weeks and have numerous associated documents, as well several that appear as emerging topics with fewer documents. All the tools and data are publicly available, and this framework can be applied to any literature collection. Taken together, these analyses produce a comprehensive, synthesized view of COVID-19 research to facilitate knowledge discovery from literature.

This study proposes a text similarity model to help biocuration efforts of the Conserved Domain Database (CDD). CDD is a curated resource that catalogs annotated multiple sequence alignment models for ancient domains and full-length proteins. These models allow for fast searching and quick identification of conserved motifs in protein sequences via Reverse PSI-BLAST. In addition, CDD curators prepare summaries detailing the function of these conserved domains and specific protein families, based on published peer-reviewed articles. To facilitate information access for database users, it is desirable to specifically identify the referenced articles that support the assertions of curator-composed sentences. Moreover, CDD curators desire an alert system that scans the newly published literature and proposes related articles of relevance to the existing CDD records. Our approach to address these needs is a text similarity method that automatically maps a curator-written statement to candidate sentences extracted from the list of referenced articles, as well as the articles in the PubMed Central database. To evaluate this proposal, we paired CDD description sentences with the top 10 matching sentences from the literature, which were given to curators for review. Through this exercise, we discovered that we were able to map the articles in the reference list to the CDD description statements with an accuracy of 77%. In the dataset that was reviewed by curators, we were able to successfully provide references for 86% of the curator statements. In addition, we suggested new articles for curator review, which were accepted by curators to be added into the reference list at an acceptance rate of 50%. Through this process, we developed a substantial corpus of similar sentences from biomedical articles on protein sequence, structure and function research, which constitute the CDD text similarity corpus. This corpus contains 5159 sentence pairs judged for their similarity on a scale from 1 (low) to 5 (high) doubly annotated by four CDD curators. Curator-assigned similarity scores have a Pearson correlation coefficient of 0.70 and an inter-annotator agreement of 85%. To date, this is the largest biomedical text similarity resource that has been manually judged, evaluated and made publicly available to the community to foster research and development of text similarity algorithms.

Capturing the semantic similarity between sentences plays a vital role in several primary applications in biomedical and clinical domains: biomedical sentence search, evidence attribution, question-answering and text summarization. In this pilot study, we evaluated the effectiveness of five representative sentence similarity models, ranging from traditional machine learning methods to the latest bidirectional transformers in the clinical domain. The evaluation was performed on a dataset consisting of over 1K sentence pairs from EMRs - the largest public dataset in this domain by far. The results show that embeddings on large biomedical corpora are the most effective methods. It also demonstrates that CNN and BERT are effective to capture sentence similarity under relatively small datasets.