SEARCH · PubMed Health

Results for “multimodal modeling”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

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

At least 181 records · Page 10Linked to original sources

PubMed1995

Interpretation of multimodal medical images using connectionist and variational methods.

Our medical objective is to match multimodal 3D medical images into a coherent model of the patient, from which diagnosis can be assessed and therapeutics guided. 3D image segmentation is absolutely necessary to reach this objective. We investigate two complementary approaches for segmenting 3D medical images. First we present some definitions, basic properties and recent theoretical results about formal neural networks, and show that these results can be applied to brain tumour segmentation. A variational approach (called the 'snake spline' method) is then detailed. We finally show how segmented 3D images can be used for multimodal image matching.

Algorithms↗

PubMed2000

HIV/AIDS-related pain as a chronic pain condition: implications of a biopsychosocial model for comprehensive assessment and effective management.

This paper reviews the current literature focusing on pain in HIV/AIDS, including prevalence, pathophysiology, substance abuse, treatment issues, and psychosocial contributions. In light of the high prevalence of pain among individuals with HIV/AIDS, attention is paid to the negative psychosocial impacts of pain in this population and to psychosocial barriers to optimal HIV/AIDS-related pain treatment. The paper conceptualizes HIV/AIDS pain as chronic pain. Subsequently, a biopsychosocial model of chronic pain assessment and treatment is applied. A multidimensional framework is presented for appropriate assessment and treatment of HIV/AIDS patients with pain, and specific recommendations and guidelines are offered for assessment and multimodal treatment of HIV/AIDS-related pain informed by the model.

Journal Article↗

PubMed2026

mmContext: an open framework for multimodal contrastive learning of omics and text data.

SUMMARY: Multimodal approaches are increasingly leveraged for integrating omics data with textual biological knowledge. Yet there is still no accessible, standardized framework that enables systematic comparison of omics representations with different text encoders within a unified workflow. We present mmContext, a lightweight and extensible multimodal embedding framework built on top of the open-source Sentence Transformers library. The software allows researchers to train or apply models that jointly embed omics and text data using any numeric representation stored in an AnnData.obsm layer and any text encoder available in Hugging Face. mmContext supports integration of diverse biological text sources and provides pipelines for training, evaluation, and data preparation. We train and evaluate models for a RNA-Seq and text integration task, and demonstrate their utility through zero-shot classification of cell types and diseases across four independent datasets. By releasing all models, datasets, and tutorials openly, mmContext enables reproducible and accessible multimodal learning for omics-text integration. AVAILABILITY AND IMPLEMENTATION: Pretrained checkpoints and full source code for our custom MMContextEncoder are available on Hugging Face huggingface.co/jo-mengr. The Python package github.com/mengerj/mmcontext provides the model implementation and training and evaluation scripts for custom training. The releases for the publication can be accessed via zenodo: adata_hf_datasets: doi.org/10.5281/zenodo.19185217 and mmContext: doi.org/10.5281/zenodo.19185493.

Computational Biology↗

PubMed2004

Circular distributions based on nonnegative trigonometric sums.

A new family of distributions for circular random variables is proposed. It is based on nonnegative trigonometric sums and can be used to model data sets which present skewness and/or multimodality. In this family of distributions, the trigonometric moments are easily expressed in terms of the parameters of the distribution. The proposed family is applied to two data sets, one related with the directions taken by ants and the other with the directions taken by turtles, to compare their goodness of fit versus common distributions used in the literature.

Animals↗

PubMed2026

Beyond predictive performance: A systematic review and critical methodological appraisal of AI/ML and conventional modelling strategies in breast, colorectal, and pancreatic Cancer.

BACKGROUND: Predictive modelling for cancer risk, treatment-related complications, and survival is central to precision oncology. Conventional logistic regression (LR) and Cox proportional hazards (CoxPH) regression remain widely used but are limited when modelling nonlinear interactions, high-dimensional imaging features, and multimodal clinical-metabolic predictors. Artificial intelligence (AI) and machine learning (ML) methods offer expanded capability through automated feature extraction, ensemble learning, and flexible survival modelling, but the evidence on when AI/ML adds value over conventional models across cancer sites and predictive tasks remains fragmented. OBJECTIVE: To systematically evaluate the methodological performance, validation strategies, and translational limitations of AI/ML models compared with conventional statistical models in published predictive-modelling studies for breast, colorectal, or pancreatic cancer. METHODS: PubMed, Scopus, and Web of Science were searched for studies published between January 2019 and March 2025. Two reviewers independently conducted title-and-abstract screening, full-text eligibility assessment, and PROBAST risk-of-bias assessment. Sixty-five studies (n = 907,567 participants) were narratively synthesised by cancer site, predictive task, model family, comparator, validation strategy, predictor modality, and calibration or explainability reporting. RESULTS: The 65 studies comprised breast cancer (n = 35), colorectal cancer (n = 21), and pancreatic cancer (n = 9). AI/ML superiority over LR and CoxPH was task- and data-dependent. CNN- and U-Net-based models predominated in imaging and body-composition tasks, tree-based ensembles consistently outperformed LR for tabular perioperative complication prediction, and CoxPH remained competitive, and in the largest pancreatic risk study, superior to XGBoost (C-index 0.802 vs 0.723) in well-structured datasets. PROBAST analysis-domain risk was moderate in 54 of 65 studies (83%), driven by limited external validation, sparse calibration reporting (11/65), and few decision-curve analyses (7/65). CONCLUSION: AI/ML adds the most methodological value in imaging-derived feature extraction and nonlinear perioperative prediction, while conventional regression remains preferable in large, structured datasets with linear predictors. Clinical translation requires standardised body-composition definitions, external validation, calibration assessment, decision-curve analysis, and explainability, in line with TRIPOD+AI and CLAIM standards.

Humans↗

PubMed1983

A systemic-developmental model of day treatment for young adult chronic patients.

The young adult chronic patient is a recently designated dysfunctional group which is growing significantly in numbers. These patients' multiple deficits in independent functioning and their extreme dependence-independence conflict make them difficult to engage and help in traditional forms of outpatient therapy. A day-treatment program which systematically and flexibily intervenes over many levels would seem to increase the likelihood of successful treatment. F. S. DuBois Day Treatment Center, Stamford, Connecticut, has established such a program based on an integration of general systems theory and separation-individuation theory. This systemic-developmental model of day treatment involves both systems-oriented multimodality therapy and developmentally phased expectations for gradually increasing independent functioning. This paper presents theory and applications of the systemic-developmental day-treatment model and a 2-year pilot program evaluation demonstrating its effectiveness.

Adult↗

PubMed2005

A gamma mixture model better accounts for among site rate heterogeneity.

MOTIVATION: Variation of substitution rates across nucleotide and amino acid sites has long been recognized as a characteristic of molecular sequence evolution. Evolutionary models that account for this rate heterogeneity usually use a gamma density function to model the rate distribution across sites. This density function, however, may not fit real datasets, especially when there is a multimodal distribution of rates. Here, we present a novel evolutionary model based on a mixture of gamma density functions. This model better describes the among-site rate variation characteristic of molecular sequence evolution. The use of this model may improve the accuracy of various phylogenetic methods, such as reconstructing phylogenetic trees, dating divergence events, inferring ancestral sequences and detecting conserved sites in proteins. RESULTS: Using diverse sets of protein sequences we show that the gamma mixture model better describes the stochastic process underlying protein evolution. We show that the proposed gamma mixture model fits protein datasets significantly better than the single-gamma model in 9 out of 10 datasets tested. We further show that using the gamma mixture model improves the accuracy of model-based prediction of conserved residues in proteins. AVAILABILITY: C++ source codes are available from the authors upon request.

Chromosome Mapping↗

PubMed2005

Multimode lasing in two-dimensional fully chaotic cavity lasers.

Multimode lasing in a fully chaotic cavity is investigated numerically by using a nonlinear dynamics model. We report a transition process from single-mode lasing to multimode lasing and reveal interactions among the lasing modes. In particular, both mode-pulling and mode-pushing interactions are shown to decrease the number of effective lasing modes. In addition, coexistence of different types of attractors of the final lasing states is numerically confirmed.

Journal Article↗

PubMed1997

Use of a stochastic simulation model to identify an efficient protocol for ovarian cancer screening.

The intervention protocol for an ovarian cancer screening trial should be efficient as well as effective, because it may become the standard of care if the trial demonstrates mortality reduction. To identify an efficient ovarian cancer screening protocol, the effectiveness and cost-effectiveness of selected single modality and multimodal screening strategies were estimated using a stochastic simulation model. Screening was simulated over a 30-year period in a hypothetical cohort of 1 million women aged 50 at the beginning of the period. The net present value of the cost per year of life saved was estimated for six protocols involving transvaginal sonography (TVS) and/or the tumor antigen CA 125. Internal and external validation was performed, and sensitivity analyses were conducted to assess the robustness of the ranking of the strategies. A multimodal strategy involving CA 125 with a threshold for positivity of either elevation above 35 U/ml or doubling since the previous screen, followed by TVS only if CA 125 is positive, was found to be efficient in the sense that no other strategies saved as many years of life at lower cost per year of life saved. Used annually, this strategy cost under $100,000 per year of life saved over a range of assumptions. The model's predictions are consistent with results reported in the literature regarding the performance of TVS and CA 125. The multimodal strategy used annually or every six months was efficient compared to either ultrasound or CA 125 used alone, over a range of assumptions. Simulation of screening may be useful in selecting a screening protocol to be tested in a randomized controlled trial.

Aged↗

PubMed2025

Q RadFusion: Hybrid Quantum Classical Radiogenomic Framework for Breast Cancer Diagnosis.

BACKGROUND AND PURPOSE: Breast cancer remains the most common cancer in women worldwide, with early and accurate diagnosis critical for patient survival. Radiogenomics integrates imaging phenotypes with genomic profiles, offering a pathway to precision diagnostics. However, existing classical machine learning models often struggle with the high dimensionality and heterogeneity of multimodal data, leading to issues in calibration and reproducibility. This study presents Q RadFusion, a hybrid quantum-classical framework designed to enhance breast cancer diagnosis by fusing mammography and genomics data. METHODS: Q RadFusion was implemented on two publicly available datasets: CBIS-DDSM (2,600 curated mammography cases, TCIA) and TCGA-BRCA (1,000 genomic profiles, GDC). Imaging preprocessing included bias-field correction, segmentation, and harmonization, while genomic data underwent normalization and imputation. Feature selection was performed using the Quantum Approximate Optimization Algorithm (QAOA), and features were mapped into a quantum Hilbert space using Variational Quantum Circuits (VQC). For multimodal fusion, ResNet encoded mammography features, and a Transformer encoded genomic features. Patient-level and site-held-out splits were used for evaluation. RESULTS: Q RadFusion achieved an AUC of 0.96 and accuracy of 94%, outperforming baselines including CNN-LSTM, ResNet + XGBoost, and multimodal Transformers. Ablation studies confirmed the contribution of quantum components, with optimal performance observed at circuit depth, qubits, and QAOA layers. The model also demonstrated improved calibration and ~ 80% fewer parameters compared to deep fusion networks. CONCLUSION: Q RadFusion demonstrates that hybrid quantum-classical radiogenomic integration can deliver accurate, reproducible, and clinically meaningful diagnostic support for breast cancer, with strong potential for future clinical translation.

Breast Cancer↗

PubMed2002

Real-time computing without stable states: a new framework for neural computation based on perturbations.

A key challenge for neural modeling is to explain how a continuous stream of multimodal input from a rapidly changing environment can be processed by stereotypical recurrent circuits of integrate-and-fire neurons in real time. We propose a new computational model for real-time computing on time-varying input that provides an alternative to paradigms based on Turing machines or attractor neural networks. It does not require a task-dependent construction of neural circuits. Instead, it is based on principles of high-dimensional dynamical systems in combination with statistical learning theory and can be implemented on generic evolved or found recurrent circuitry. It is shown that the inherent transient dynamics of the high-dimensional dynamical system formed by a sufficiently large and heterogeneous neural circuit may serve as universal analog fading memory. Readout neurons can learn to extract in real time from the current state of such recurrent neural circuit information about current and past inputs that may be needed for diverse tasks. Stable internal states are not required for giving a stable output, since transient internal states can be transformed by readout neurons into stable target outputs due to the high dimensionality of the dynamical system. Our approach is based on a rigorous computational model, the liquid state machine, that, unlike Turing machines, does not require sequential transitions between well-defined discrete internal states. It is supported, as the Turing machine is, by rigorous mathematical results that predict universal computational power under idealized conditions, but for the biologically more realistic scenario of real-time processing of time-varying inputs. Our approach provides new perspectives for the interpretation of neural coding, the design of experiments and data analysis in neurophysiology, and the solution of problems in robotics and neurotechnology.

Action Potentials↗

PubMed2005

A model for transits in dynamic response theory.

The first goal of vibration-transit (V-T) theory was to construct a tractable approximate Hamiltonian from which the equilibrium thermodynamic properties of monatomic liquids can be calculated. The Hamiltonian for vibrations in an infinitely extended harmonic random valley, together with the universal multiplicity of such valleys, gives an accurate first-principles account of the measured thermodynamic properties of the elemental liquids at melt. In the present paper, V-T theory is extended to nonequilibrium properties, through an application to the dynamic structure factor S(q,omega). It was previously shown that the vibrational contribution alone accurately accounts for the Brillouin peak dispersion curve for liquid sodium, as compared both with molecular-dynamics (MD) calculations and inelastic x-ray scattering data. Here it is argued that the major effects of transits will be to disrupt correlations within the normal-mode vibrational motion and to provide an additional source of inelastic scattering. We construct a parametrized model for these effects and show that it is capable of fitting MD results for S(q,omega) in liquid sodium. A small discrepancy between model and MD at large q is attributed to multimode vibrational scattering. In comparison, mode coupling theory formulates S(q,omega) in terms of processes through which density fluctuations decay. While mode coupling theory is also capable of modeling S(q,omega) very well, V-T theory is the more universal since it expresses all statistical averages, thermodynamic functions, and time correlation functions alike, in terms of the same motional constituents, vibrations and transits.

Journal Article↗

PubMed2002

[Principles on therapy of myoarthropathic pain].

The myoarthropathic pain is the major source of orofacial pain. The diagnosis is not always easy because the diagnostic signs are not pathognomonic and they may occur also with other pain disorders. The pain intensity fluctuates, and mild to medium intensity pain has the tendency to subside spontaneously or can be alleviated with simple, non-invasive therapies as counseling, self-control, analgesics, physiotherapy, and occlusal appliances. For most patients a combination of counseling, self-control and physiotherapy (home program) is sufficient to relieve the pain. Occlusal appliances should therefore be used only if these modalities fail or in presence of pain on awakening, ostheoarthritis, a painful discopathy, and/or a malocclusion caused by condylar remodeling secondary to ostheoarthrosis. Only in a small percentage of patients the myoarthropathic pain persists and becomes chronic. These patients need a multimodal therapy, according to the biopsychosocial pain model. In addition to the modalities just described, the treatment must address also the affective, emotional, cognitive, and behavioral pain component and be tailored to the single patient based on his/her psychosocial and constitutional characteristics. The goal is not pain relieve but improvement of the quality of life by teaching the patient more efficient pain coping strategies by means of behavioral and relaxation therapy. In addition, tricyclic antidepressants may be used in order to treat the somatic pain component caused by the neuroplastic changes that take place in the central nervous system in chronic pain conditions.

Affect↗

PubMed2003

[Muscular-arthroscopic pain: often unappreciated].

Myoarthropathic pain, most often due to masticatory myalgia, is the major source of orofacial pain. Its diagnosis is not always easy because the diagnostic signs are not pathognomonic and they may occur also with other pain disorders. The pain intensity fluctuates, and mild to medium intensity pain has the tendency to subside spontaneously or can be alleviated with simple, non-invasive therapies such as counseling, self-control, analgesics, physiotherapy, and occlusal appliances. For most patients a combination of counseling, self-control and physiotherapy (home program) is sufficient to relieve the pain. Occlusal appliances should, therefore, be used only if these modalities fail or if pain is present on awakening. Only in a small percentage of patients the myoarthropathic pain persists and becomes chronic. The factors that determine pain chronicity are complex and most likely due to the interaction of biomedical factors - such as the persistence of peripheral nociceptive input, neuroplastic changes at spinal and supraspinal levels (cortical reorganisation)- with psychosocial factors, in addition, possibly, to a polymorphism of the catechol-o-methyltransferase gene. Patients with chronic myogenic pain need a multimodal therapy, according to the biopsychosocial pain model. In addition to the modalities just described, the treatment must address also the affective, emotional, cognitive and behavioural pain component and be tailored to the single patient based on his/her psychosocial and constitutional characteristics. The goal is not pain relief but improvement of the quality of life by teaching the patient more efficient pain coping strategies by means of behavioural and relaxation techniques. In addition, tricyclic antidepressants may be used in order to treat the somatic pain component caused by the neuroplastic changes that take place in the central nervous system in chronic pain conditions.

Craniomandibular Disorders↗

PubMed2005

Testing models of thalamic dysfunction in schizophrenia using neuroimaging.

Neural models of schizophrenia have implicated the thalamus in deficits of early sensory processing and multimodal integration. We have reviewed the existing neuroimaging literature for evidence in support of models that propose abnormalities of thalamic relay nuclei, the mediodorsal thalamic nucleus, and large-scale cortico-thalamic networks. Thalamic volume reduction was found in some but not all studies. Studies of the early stages of schizophrenia suggest that thalamic volume reduction is present early in the course of the illness. Functional imaging studies have revealed task related abnormalities in several cortical and subcortical areas including the thalamus, suggesting a disruption of distributed thalamocortical networks. Chemical imaging studies have provided evidence for a loss of thalamic neuronal integrity in schizophrenia. There is, at present, inadequate data to support the hypothesis that schizophrenia is associated with abnormalities of sensory relay or association nuclei. There is evidence for a perturbation of cortico-thalamic networks, but further research is needed to elucidate the underlying mechanisms at the cellular and systems levels. The challenges ahead include better delineation of thalamic structure and function in vivo, the combination of genetic and imaging techniques to elucidate the genetic contributions to a thalamic phenotype of schizophrenia, and longitudinal studies of thalamic structure and function.

Animals↗

PubMed2006

Identification and analysis of multimode guided waves in tibia cortical bone.

The hollow cylinder model filled with viscous-liquid is performed to simulate tibial bone shape in this paper. Short time Fourier transform (STFT) was applied to identify and analyze guided wave modes of a series of ultrasonic signals acquired at various transmitter-receiver positions for a fixed transmitter. Experimental results were compared with theoretical simulation. The high correlation values for group velocity were obtained between results of experiment and theory for L(0,2) (r=0.89, p<10(-5)), and L(0,3) (r=0.87, p<10(-5)). The experimental group velocities correlated strongly with cortical thickness (cortTh) for L(0,2) (r(2)=0.79, p<10(-5)) and L(0,3) (r(2)=0.74, p<10(-5)) at 0.5 MHz center frequency. The results showed that the STFT is an effective method to identify the propagating modes and derive dispersion information. The experimental results were in good agreement with theoretical predictions. This study suggests that the guided wave L(0,2) mode is a promising mode to assess change in cortical thickness.

Algorithms↗

PubMed2006

The yips in golf: multimodal evidence for two subtypes.

PURPOSE: To determine whether a model of two subtypes of yips is supported by evidence from a range of physiological, behavioral, and psychological measures. METHODS: Fifteen golfers who experience yips symptoms while putting (mean age 58.1 yr, SD 13.6 yr), and nine golfers with no yips symptoms (mean age 39.6 yr, SD 19.3 yr) were recruited. Participants completed a golf history questionnaire to determine their playing experience and the nature of any yips symptoms experienced. In experiment 1, participants performed a putting task while electromyographic data were recorded from the forearm flexors and extensors and biceps brachii, bilaterally. The task was performed in two sessions, under low-pressure and high-pressure experimental conditions. The high-pressure condition was intended to increase anxiety through the use of a monetary incentive, video-taping of performance, and the presence of a confederate who provided negative feedback. Participants' state of anxiety was assessed using a questionnaire before each of the experimental sessions. In experiment 2, participants completed a task that required the inhibition of an anticipated response. Their accuracy and ability to inhibit their response was determined. RESULTS: The golfers who experienced yips could be categorized according to whether they reported mainly movement-related symptoms (Type I) or anxiety-related symptoms (Type II). The Type I group exhibited greater muscle activity during putting and greater errors and less inhibition of the anticipated response task. The Type II group exhibited greater changes in cognitive anxiety and normal performance of the anticipated response task. CONCLUSION: This study provides evidence in support of two yips subtypes. Type I is related to impaired movement initiation and execution, whereas Type II is related to performance anxiety.

Adolescent↗

PubMed2006

Scattering coherent and partially coherent space-time pulses through few-mode optical systems.

We consider the spatiotemporal behavior of coherent and partially coherent, pulsed, few-mode optical systems. It is shown that there is some set of orthogonal space-time pulses at the input reference surface that maps in one-to-one correspondence with some set of orthogonal space-time pulses at the output reference surface; we call these pulses eigenfields. The spectrum of the coupling coefficients determines the amount of information that can be transmitted within a given period of time. The eigenfields are unique for a given system and can be used to propagate a field that is in any state of spatial and temporal coherence. They can also be used to account for the spatial and temporal coherence of internally generated noise and to calculate the powers, fluctuations, and correlations that would be recorded by multimode detectors. Our technique is ideal for modeling the behavior of pulsed imaging arrays and interferometers.

Journal Article↗