Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Foundation model”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 55 records · Page 3Linked to original sources
PubMed2003

Logic-based remodeling of the Digital Anatomist Foundational Model.

This paper describes a development cycle for the engineering of large knowledge bases: A graphical tool is used for editing and the content is transformed into a logic-based representation language. This representation is used to check the consistency of the knowledge base as well as to facilitate the reviewing process. Showing the usefulness of this approach, aspects of the Digital Anatomist Foundational Model will be transformed into a Description Logics representation. We introduce a special modeling technique to account for the representation of the complex part/whole relationships in the biomedical domain.

Anatomy↗
PubMed2025

Foundation model based multimodal transformer framework for survival analysis in HER2 stratified breast cancer.

Objective. To improve survival prediction for HER2-positive breast cancer by integrating histopathological, molecular, and clinical data using a multimodal transformer framework.Approach. We propose a multimodal transformer framework for breast cancer survival prediction using HER2 stratified (SurvMBC), a foundation model-enhanced architecture that fuses three data modalities: whole-slide images, clinical narratives, and molecular features. Tumor microenvironment features are extracted using a pathology language and image pre-training (PLIP), clinical narratives are processed with BioBERT, and miRNA expression plus DNA methylation data are embedded using Gen2Vec. These representations are integrated through a cross-modal transformer with attention mechanisms for survival prediction.Main results. The model was evaluated on 1,095 HER2-positive breast cancer patients from The Cancer Genome Atlas. SurvMBC achieved a concordance index (C-index) of 0.857 (95% CI: 0.834, 0.880), a low integrated Brier score, and a strong inverse negative binomial log-likelihood. Risk stratification based on model outputs significantly separated high- and low-risk groups (log-rankp< 0.01) and showed strong associations with tumor stage, grade, and hormone receptor status (allp< 0.05).Significance. SurvMBC demonstrates the effectiveness of multimodal fusion in addressing tumor heterogeneity and improving prognostic accuracy. The attention-based integration enables context-aware learning of survival-relevant features across modalities, supporting individualized risk stratification and risk-adaptive treatment planning for HER2 stratified breast cancer patients.

Breast Neoplasms↗
PubMed2006

The WHO-ILAR COPCORD Bhigwan (India) model: foundation for a future COPCORD design and data repository.

Launched by the International League of Associations for Rheumatology (ILAR) and the World Health Organization (WHO), the Community oriented program for control of rheumatic diseases (COPCORD) aims to fill the gaps in the knowledge on the global burden of rheumatic musculoskeletal disorders (RMS). During the population survey (Stage I), data on symptoms (pain and disability in focus), rather than diseases or syndromes, is collected. The survey may be followed by a planned stage to impart health education, identify risk factors, and devise preventive and control strategies. Several countries in the Asia Pacific and Pan-America have completed COPCORD survey. Africa has recently joined. Only COPCORD Bhigwan (India) has continued into the tenth year. COPCORD Bhigwan is a fast-track model that has provided significant data on rheumatic disorders. Using COPCORD Bhigwan model, the Bone and Joint Decade (BJD) India has launched several population surveys to measure the RMS burden. There is an urgent need for a COPCORD data repository. Several COPCORD have differed in their methods. Differences pertain to population sample size, techniques for data collection and recording, chronology of events and phases, and classification of symptoms/diseases/disorders. The COPCORD model in current global use needs to be revised. Based on the COPCORD Bhigwan model, a future design for COPCORD is proposed. COPCORD needs to have a uniform and standardized core program with a flexibility to cater to regional needs. It must imbibe some of the recent advances in rheumatology while retaining its socioeconomic appeal. It must have a planned follow-up/longitudinal observational phase. Above all, it must serve and benefit community. WHO-ILAR COPCORD and the global BJD initiative must join hands to serve a common cause of controlling rheumatic musculoskeletal disorders. COPCORD is also a reflection of the ILAR mission statement "think global, act local."

Community Health Planning↗
PubMed2026

GBFN: A gated bimodal fusion network leveraging foundation model embeddings for cancer drug sensitivity prediction.

Despite recent progress in deep learning for cancer drug sensitivity prediction, many existing models still rely on task-specific representation learning or relatively simple multimodal fusion, which may limit their ability to capture complex drug-cell interactions. To address this issue, we developed GBFN, a gated bimodal fusion network for continuous IC50 prediction that integrates pretrained drug and cell-line representations. Specifically, drug embeddings were obtained from SMI-TED, whereas cell-line embeddings were derived from transcriptomic profiles using BulkFormer. These two modalities were then combined through a dimension-wise gated fusion module and used to predict IC50 values in matched drug-cell line pairs. On the CCLE-based benchmark, GBFN outperformed representative neural baselines, including GraphDRP, TGSA, and TransEDRP, and achieved the best overall performance, with an R&#xb2; of 0.8714 and an RMSE of 0.8938. Moreover, ablation analysis showed that the model using drug features and cell-line expression data with gated fusion performed better than the corresponding model using direct concatenation, indicating that the improvement was associated with the fusion strategy rather than with the input modalities alone. In addition, cell-line expression data were more informative than mutation data in the present setting, and adding mutation data to the model using drug features and expression data did not further improve performance. Across major cancer types, GBFN maintained generally high cell-line-level predictive performance, and perturbation-based attribution identified biologically relevant transcriptomic programs in selected drug-cell line settings. Together, these findings support GBFN as a compact and effective framework for continuous drug response prediction.

Humans↗
PubMed1999

Influence of the Digital Anatomist Foundational Model on traditional representations of anatomical concepts.

A principled and logical representation of the structure of the human body has led to conflicts with traditional representations of the same knowledge by anatomy textbooks. The examples which illustrate resolution of these conflicts suggest that stricter requirements must be met for semantic consistency, expressivity and specificity by knowledge sources intended to support inference than by textbooks and term lists. These next-generation resources should influence traditional concept representation, rather than be constrained by convention.

Anatomy↗
PubMed2006

Restructuring the foundational model of anatomy.

The authors present a method to convert the FMA to a description logic-based representation in OWL. The concepts denoting anatomical structures are aligned to the DOLCE formal top-level ontology, and converted to a compact core ontology in the spirit of GALEN. The paper presents the identified problems in the FMA and the main aspects of the re-modelling.

Humans↗
PubMed1996

Tumor heterogeneity and progression: conceptual foundations for modeling.

A conceptual foundation for modeling tumor progression, growth, and heterogeneity is presented. The purpose of such models is to aid understanding, test ideas, formulate experiments, and to model cancer 'in machina' to address the dynamic features of tumor cell heterogeneity, progression, and growth. The descriptive capabilities of such an approach provides a consistent language for qualitatively reasoning about tumor behavior. This approach provides a schema for building conceptual models that combine three key phenomenological driving elements: growth, progression, and genetic instability. The growth element encompasses processes contributing to changes in tumor bulk and is distinct from progression per se. The progression element subsumes a broad collection of processes underlying phenotypic progression. The genetics elements represents heritable changes which potentially affect tumor character and behavior. Models, conceptual and mathematical, can be built for different tumor situations by drawing upon the interaction of these three distinct driving elements. These models can be used as tools to explore a diversity of hypotheses concerning dynamic changes in cellular populations during tumor progression, including the generation of intratumor heterogeneity. Such models can also serve to guide experimentation and to gain insight into dynamic aspects of complex tumor behavior.

Animals↗
PubMed2000

Prediction of pesticide concentrations in the atmosphere using an atmospheric diffusion model (linear source plume model).

The foundational model to predict concentration of pesticides in the atmosphere outside of the sprayed area was developed using the results of measured concentration in the atmosphere, in reference to the atmospheric diffusion model utilized for the air pollution prediction model. The atmospheric diffusion model assumes that the applied area was a topographically flat farmland, that wind direction and wind speeds were constant, and the pesticide was constantly discharged from the emission line sources. Therefore the linear source plume model (LSPLM) was developed. The concentration in the atmosphere was predicted by assigning the property of the pesticides and various conditions of measurements of the model, and compared with the measured them, then the adaptability of the model was examined. As a result, the correlation between the measured value and the predicted value in paddy and forested areas was significant (P < 0.01) although deviations in the order of tens were observed, the measured value and the predicted value were generally in agreement.

Agriculture↗
PubMed2026

Large language models in bioinformatics: a comprehensive survey.

The emergence of foundation models with trillion-level parameters has redefined the landscape of artificial intelligence. Various fields are developing their own large-scale models, which can solve many problems within the field and improve work efficiency. Biological large-scale models are a cross-disciplinary research field that combines mathematics, computer science, and biology, aiming to simulate and understand the structure, function, and dynamic changes of biological systems through the establishment of complex computational models. This field covers multiple levels such as biological pathways, population dynamics, protein folding, etc., providing us with tools for deep exploration of the mysteries of life and applications in medicine, ecology, and other fields. This article reviews the background and research status of biological large-scale models, and discusses future directions. Large language models (LLMs) and other large-scale foundation models have rapidly advanced in recent years, enabling powerful representation learning and generation across text, sequences, and multimodal data. In bioinformatics and biomedicine, these models are increasingly used to analyze genomic sequences, infer protein properties and structures, support drug discovery, and integrate heterogeneous biomedical evidence. This survey reviews the basic principles of LLMs and summarizes representative applications in (i) gene and genome sequence analysis, (ii) protein structure and function prediction, and (iii) drug design, including virtual screening and personalized medicine. We also discuss emerging multi-model modeling approaches, as well as key challenges such as data quality and privacy, interpretability, generalization to new organisms and tasks, and responsible deployment in health-related settings. Finally, we outline future directions for developing reliable, scalable, and explainable bioinformatics foundation models.

bioinformatics↗
PubMedDate not supplied

A three-dimensional, anatomically detailed foot model: a foundation for a finite element simulation and means of quantifying foot-bone position.

We generated an anatomically detailed, three-dimensional (3-D) reconstruction of a human foot from 286 computerized topographic (CT) images. For each bone, 2-D cross-sectional data were obtained and aligned to form a stacked image model. We calculated the inertial matrix of each bone from the stacked image model and used it to determine the principal axes. Relative angles between the principal axes of the bones were employed to describe the shape of the foot, i.e., the relationships between the bones of the foot. A 3-D surface model was generated from the stacked image models and a detailed 3-D mesh for each bone was created. Additionally, the representative geometry of the plantar soft tissue was obtained from the CT scans, while the geometries of the cartilage between bones were obtained from the 3-D surface bone models. This model served dual purposes: it formed the anatomical foundation for a future finite element model of the human foot and we used it to objectively quantify foot shape using the relationships between the principal axes of the foot bones.

Aged↗
PubMedDate not supplied

Predicting the outcome of renal replacement therapy in severe acute renal failure.

Continuous renal replacement therapy (CRRT), such as continuous venovenous hemofiltration, has theoretical advantages over intermittent hemodialysis (IHD) that are related to cardiorespiratory stability, metabolic control, and fluid balance allowing nutritional supplementation. However, retrospective and controlled studies fail to show these advantages because of comorbidity associated with triage to CRRT. To compare outcomes using IHD versus CRRT, we applied published risk stratification models (Cleveland Clinic Foundation, Lohr index, and APACHE II) to the 349 patients with acute renal failure requiring renal replacement therapy at University of Michigan over the 2 year period including 1995 and 1996. The Cleveland Clinic Foundation model best predicted overall mortality, but our CRRT patients had excess, unpredicted mortality that was particularly prominent in the lower risk categories. The Lohr clinical score predicted mortality less accurately but also was associated with higher, unpredicted mortality at lower risk scores among the CRRT patients. APACHE II scores did not predict mortality very well among IHD, CRRT, or the combined group of patients. We conclude that the need for CRRT itself predicts mortality over and above that included in published risk models. Either CRRT is associated with some unidentified morbidity (e.g., treatment associated infection) or, more likely, triage to CRRT is associated with as yet unspecified comorbidity not detected in existing risk stratification schemes. It will be important to address these issues in any future studies evaluating outcome or comparing renal replacement therapy modalities among patients with severe acute renal failure.

APACHE↗
PubMed1991

Application of beams on elastic foundation and B-spline solution methodologies to parametric analysis of intramedullary implant systems.

A simple numerical technique for parametric evaluation of orthopaedic implant systems, to be used as a screening tool before complex structural analysis (e.g. Finite Element Method), is the subject of this paper. A modified Beams on Elastic Foundation model (with non-constant foundation modulus) is solved using this numerical technique based on B-spline differential equation modelling. A model with variation in the modulus of the foundation, as solved with this spline technique, was compared with a model with constant foundation modulus, solvable with closed form techniques. While deflections were smaller, the reaction force was up to ten times greater for the models with constant modulus of foundation, compared with varying modulus. The model presented in this paper is a refinement of previous models using closed form solution techniques for foundations with constant moduli. It is primarily useful for detecting trends in parametric analyses, or to select specific cases for further analysis by more computationally intensive analytic methods.

Biomechanical Phenomena↗