Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “probabilistic modelling”

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 1,081 records · Page 60Linked to original sources
PubMed2005

Probabilistic segmentation and intensity estimation for microarray images.

We describe a probabilistic approach to simultaneous image segmentation and intensity estimation for complementary DNA microarray experiments. The approach overcomes several limitations of existing methods. In particular, it (a) uses a flexible Markov random field approach to segmentation that allows for a wider range of spot shapes than existing methods, including relatively common 'doughnut-shaped' spots; (b) models the image directly as background plus hybridization intensity, and estimates the two quantities simultaneously, avoiding the common logical error that estimates of foreground may be less than those of the corresponding background if the two are estimated separately; and (c) uses a probabilistic modeling approach to simultaneously perform segmentation and intensity estimation, and to compute spot quality measures. We describe two approaches to parameter estimation: a fast algorithm, based on the expectation-maximization and the iterated conditional modes algorithms, and a fully Bayesian framework. These approaches produce comparable results, and both appear to offer some advantages over other methods. We use an HIV experiment to compare our approach to two commercial software products: Spot and Arrayvision.

Algorithms↗
PubMedDate not supplied

A neural network approach to approximating MAP in belief networks.

Bayesian belief networks (BBN) are a widely studied graphical model for representing uncertainty and probabilistic interdependence among variables. One of the factors that restricts the model's wide acceptance in practical applications is that the general inference with BBN is NP-hard. This is also true for the maximum a posteriori probability (MAP) problem, which is to find the most probable joint value assignment to all uninstantiated variables, given instantiation of some variables in a BBN. To circumvent the difficulty caused by MAP's computational complexity, we suggest in this paper a neural network approximation approach. With this approach, a BBN is treated as a neural network without any change or transformation of the network structure, and the node activation functions are derived based on an energy function defined over a given BBN. Three methods are developed. They are the hill-climbing style discrete method, the simulated annealing method, and the continuous method based on the mean field theory. All three methods are for BBN of general structures, with the restriction that nodes of BBN are binary variables. In addition, rules for applying these methods to noisy-or networks are also developed, which may lead to more efficient computation in some cases. These methods' convergence is analyzed, and their validity tested through a series of computer experiments with two BBN of moderate size and complexity. Although additional theoretical and empirical work is needed, the analysis and experiments suggest that this approach may lead to effective and accurate approximation for MAP problems.

Algorithms↗
PubMed2004

Measurement of the activity of 11C and 22Na sources using a 4pi-beta-gamma coincidence system.

Many radiation applications, including positron emission tomography (PET) studies and activation dosimetry, require the knowledge of the activity of short-lived radionuclide samples, whereas relative measurements may be hampered by the absence of a reference source. Using 11C radionuclide as an example, an analytical model based on a probabilistic approach has been set up to predict the activity of a pure positron emitter measured using the 4pi-beta-gamma coincidence technique. The model has been extended to describe the measurement on a 22Na source used to test the measurement technique. Comparison of the modeled results with the measurements confirms the general validity of the model. The model has also been studied for the effect of the variation of key measurement conditions, such as nominal source activity, detector efficiency, detector background levels, and coincidence resolving time. The 4pi-beta-gamma coincidence technique and the results of modeling allow the activity measurements on 22Na and 11C sources with an estimated relative standard uncertainly on the order of one percent.

Carbon Radioisotopes↗
PubMed2006

Hidden Markov model variants and their application.

Markov statistical methods may make it possible to develop an unsupervised learning process that can automatically identify genomic structure in prokaryotes in a comprehensive way. This approach is based on mutual information, probabilistic measures, hidden Markov models, and other purely statistical inputs. This approach also provides a uniquely common ground for comparative prokaryotic genomics. The approach is an on-going effort by its nature, as a multi-pass learning process, where each round is more informed than the last, and thereby allows a shift to the more powerful methods available for supervised learning at each iteration. It is envisaged that this "bootstrap" learning process will also be useful as a knowledge discovery tool. For such an ab initio prokaryotic gene-finder to work, however, it needs a mechanism to identify critical motif structure, such as those around the start of coding or start of transcription (and then, hopefully more).For eukaryotes, even with better start-of-coding identification, parsing of eukaryotic coding regions by the HMM is still limited by the HMM's single gene assumption, as evidenced by the poor performance in alternatively spliced regions. To address these complications an approach is described to expand the states in a eukaryotic gene-predictor HMM, to operate with two layers of DNA parsing. This extension from the single layer gene prediction parse is indicated after preliminary analysis of the C. elegans alt-splice statistics. State profiles have made use of a novel hash-interpolating MM (hIMM) method. A new implementation for an HMM-with-Duration is also described, with far-reaching application to gene-structure identification and analysis of channel current blockade data.

Alternative Splicing↗
PubMed2006

From single to multiple deficit models of developmental disorders.

The emerging etiological model for developmental disorders, like dyslexia, is probabilistic and multifactorial while the prevailing cognitive model has been deterministic and often focused on a single cognitive cause, such as a phonological deficit as the cause of dyslexia. So there is a potential contradiction in our explanatory frameworks for understanding developmental disorders. This paper attempts to resolve this contradiction by presenting a multiple cognitive deficit model of developmental disorders. It describes how this model evolved out of our attempts to understand two comorbidities, those between dyslexia and attention deficit hyperactivity disorder (ADHD) and between dyslexia and speech sound disorder (SSD).

Attention Deficit Disorder with Hyperactivity↗
PubMed2002

Analysing longitudinal continuous quality of life data with dropout.

Quality of Life (QL) is becoming an increasingly popular endpoint in phase III cancer clinical trials. However, there is still no agreement as to what is the optimal approach to analysis. In this paper we review some concepts which should be considered during a QL analysis. We present two modelling approaches that have been substantively developed in other research fields: selection models and pattern-mixture models. These models are compared using data from an EORTC clinical trial in poor-prognosis prostate cancer patients. It is illustrated that, although selection models and pattern mixture are probabilistically equivalent, they may shed completely different light on data from a modeller's point of view.

Clinical Trials, Phase III as Topic↗
PubMed2005

Probabilistic finite element prediction of knee wear simulator mechanics.

Computational models have recently been developed to replicate experimental conditions present in the Stanmore knee wear simulator. These finite element (FE) models, which provide a virtual platform to evaluate total knee replacement (TKR) mechanics, were validated through comparisons with experimental data for a specific implant. As with any experiment, a small amount of variability is inherently present in component alignment, loading, and environmental conditions, but this variability has not been previously incorporated in the computational models. The objectives of the current research were to assess the impact of experimental variability on predicted TKR mechanics by determining the potential envelope of joint kinematics and contact mechanics present during wear simulator loading, and to evaluate the sensitivity of the joint mechanics to the experimental parameters. In this study, 8 component alignment and 4 experimental parameters were represented as distributions and used with probabilistic methods to assess the response of the system, including interaction effects. The probabilistic FE model evaluated two levels of parameter variability (with standard deviations of component alignment parameters up to 0.5mm and 1 degrees ) and predicted a variability of up to 226% (3.44mm) in resulting anterior-posterior (AP) translation, up to 169% (4.30 degrees ) in internal-external (IE) rotation, but less than 10% (1.66MPa) in peak contact pressure. The critical alignment parameters were the tilt of the tibial insert and the IE rotational alignment of the femoral component. The observed variability in kinematics and, to a lesser extent, contact pressure, has the potential to impact wear observed experimentally.

Arthroplasty, Replacement, Knee↗
PubMed2005

Predictive risk thresholds for survival protection of farmed abalone, Haliotis diversicolor supertexta, exposed to waterborne zinc.

Using a probabilistic risk-based framework, we have developed a simple predictive risk threshold model for protecting the survival of farmed abalone, Haliotis diversicolor supertexta, exposed to waterborne zinc (Zn). Probabilistic techniques using Monte Carlo analysis propagate parameter uncertainty/variability throughout the model, providing decision makers with a credible range of information and increased flexibility in establishing a specific Zn level in aquacultural ecosystems. We coupled a first-order two-compartment bioaccumulation model with a reconstructed dose-response profile based on a three-parameter Hill equation model to form a probabilistic risk model in order to determine the risk quotient associated with a 10% probability of exceeding the abalone 5% effect concentration (EC(5)) at site-specific abalone farms. Sensitivity analysis revealed that waterborne Zn concentration (C(w)) and algae bioconcentration factor (BCF(a)) have a significant effect on Zn levels in abalone. Using multiple nonlinear regression analysis with C(w) and BCF(a) as the parameters, a predictive risk threshold equation that can be used in a variety of site-specific conditions was developed for protecting the survival of farmed abalone. We believe this probabilistic framework provides an effective method for conceptualizing a public policy decision vis-a-vis the establishment of a specific acceptable risk threshold for aquacultural water quality management.

Animals↗
PubMed2003

Backward location and travel time probabilities for a decaying contaminant in an aquifer.

Backward location and travel time probabilities can be used to determine the prior position of contamination in an aquifer. These probabilities, which are related to adjoint states of concentration, can be used to improve characterization of known sources of groundwater contamination, to identify previously unknown contamination sources, and to delineate capture zones. The first contribution of this paper is to extend the adjoint model to the case of a decaying solute (first-order decay), and to describe two different interpretations of backward probabilities. The conventional interpretation accounts for the probability that a contaminant particle could decay before reaching the detection location. The other interpretation is conditioned on the fact that the detected contaminant particle actually reached the detection location, despite this possibility of decay. In either case, travel time probabilities are skewed toward earlier travel times, relative to a conservative solute. The second contribution of this paper is to verify the load term for a monitoring well observation. We provide examples using one-dimensional models and hypothetical aquifers. We employ an infinite domain in order to verify the monitoring well load. This new but simple one-dimensional adjoint solution can also be used to verify higher-dimensional numerical models of backward location and travel time probabilities. We employ a semi-infinite domain to illustrate the effect of decay on backward models of pumping well probabilistic capture zones. Decay causes the capture zones to fall closer to the well.

Humans↗
PubMedDate not supplied

[A probabilistic approach to the natural course of liver cirrhosis: a model of interactive simulation].

We propose a computerized mathematical model of the natural history of patients with liver cirrhosis in order to investigate the relationships between the various events that characterize the clinical course of the disease. By means of a rigorously probabilistic approach, the stimulation exposes each hypothetical patient to a series of risks of significant clinical events; whether or not these events occur is determined by preestablished probabilities. Analysis of the effects of arbitrary modifications of these theoretical risks provides useful information on the prospective relevance of the clinical events examined. For example, we compared the simulated follow-up of a population of control cirrhotic subjects with follow-ups of patient groups in whom different degrees of reduction of the risk for gastroesophageal bleeding were postulated on the basis of hypothetical primary or secondary prophylaxis. Our data confirm that the size of the investigated population strongly influences the threshold level of the bleeding-risk reduction able to determine a significant improvement in overall survival. In fact, in large populations (a 6 year follow-up of 3000 cirrhotic subjects), the threshold level was 25% (primary risk) and 45% (secondary risk) with respect to control subjects, while in more limited populations (300 subjects) even a 100% risk-reduction does not seem sufficient to bring about significant improvement in overall survival. Our data may help to explain the different results of controlled trials found in the literature; in fact, only the conclusions arrived at through meta-analysis procedures in large populations have been confirmed.(ABSTRACT TRUNCATED AT 250 WORDS)

Computer Simulation↗
PubMed2005

Confidence mapping in diffusion tensor magnetic resonance imaging tractography using a bootstrap approach.

The bootstrap technique is an extremely powerful nonparametric statistical procedure for determining the uncertainty in a given statistic. However, its use in diffusion tensor MRI tractography remains virtually unexplored. This work shows how the bootstrap can be used to assign confidence to results obtained with deterministic tracking algorithms. By invoking the concept of a "tract-propagator," it also underlines the important effect of local fiber architecture or architectural milieu on tracking reproducibility. Finally, the practical advantages and limitations of the technique are discussed. Not only does the bootstrap allow any deterministic tractography algorithm to be used in a probabilistic fashion, but also its model-free inclusion of all sources of variability (including those that cannot be modeled) means that it provides the most realistic approach to probabilistic tractography.

Algorithms↗
PubMed2005

High-resolution iterative frequency identification for NMR as a general strategy for multidimensional data collection.

We describe a novel approach to the rapid collection and processing of multidimensional NMR data: "high-resolution iterative frequency identification for NMR" (HIFI-NMR). As with other reduced dimensionality approaches, HIFI-NMR collects n-dimensional data as a set of two-dimensional (2D) planes. The HIFI-NMR algorithm incorporates several innovative features. (1) Following the initial collection of two orthogonal 2D planes, tilted planes are selected adaptively, one-by-one. (2) Spectral space is analyzed in a rigorous statistical manner. (3) An online algorithm maintains a model that provides a probabilistic representation of the three-dimensional (3D) peak positions, derives the optimal angle for the next plane to be collected, and stops data collection when the addition of another plane would not improve the data model. (4) A robust statistical algorithm extracts information from the plane projections and is used to drive data collection. (5) Peak lists with associated probabilities are generated directly, without total reconstruction of the 3D spectrum; these are ready for use in subsequent assignment or structure determination steps. As a proof of principle, we have tested the approach with 3D triple-resonance experiments of the kind used to assign protein backbone and side-chain resonances. Peaks extracted automatically by HIFI-NMR, for both small and larger proteins, included approximately 98% of real peaks obtained from control experiments in which data were collected by conventional 3D methods. HIFI-NMR required about one-tenth the time for data collection and avoided subsequent data processing and peak-picking. The approach can be implemented on commercial NMR spectrometers and is extensible to higher-dimensional NMR.

Animals↗
PubMed1999

Prognostic methods in medicine.

Prognosis--the prediction of the course and outcome of disease processes--plays an important role in patient management tasks like diagnosis and treatment planning. As a result, prognostic models form an integral part of a number of systems supporting these tasks. Furthermore, prognostic models constitute instruments to evaluate the quality of health care and the consequences of health care policies by comparing predictions according to care norms with actual results. Approaches to developing prognostic models vary from using traditional probabilistic techniques, originating from the field of statistics, to more qualitative and model-based techniques, originating from the field of artificial intelligence (AI). In this paper, various approaches to constructing prognostic models, with emphasis on methods from the field of AI, are described and compared.

Artificial Intelligence↗
PubMed2003

Probabilistic assessment of contamination using the two-phase flow model.

A physically motivated model is indispensable for a successful analysis of the impact of leaching from nuclear waste storage sites on the environment and public health. While most analyses use the single-phase flow model for modelling unsaturated flow and solute transport, the two-phase flow model considering the resistance of gas to water flow is a more realistic one. The effect of the two-phase flow model on the water content is theoretically investigated first in this study. Then, by combining a geostatistical generator using the turning bands method and a multi-phase transport code TOUGH2, an automatic process is used for Monte Carlo simulation of the solute transport. This stochastic approach is applied to a potentially polluted site by low-level nuclear waste in Taiwan. In the simulation, the saturated hydraulic conductivity is treated as the random variable. The stochastic approach provides a probabilistic assessment of contamination. The results show that even though water content from the two-phase flow model is only 1.5% less than the one from the single-phase flow model, the two-phase flow causes a slower movement but a wider lateral spreading of the plume in the unsaturated zone. The stochastic approach provides useful probability information which is not available from the deterministic approach. The probability assessment of groundwater contamination provides the basis for more informed waste management, better environmental assessment and improved evaluation of impact on public health.

Forecasting↗
PubMed2006

Local modeling of the rolling phenomenon in the cell adhesion kinetics: A probabilistic approach.

Cell adhesion plays an important role in biology: essentially with regard to immunizing defence and the transport of medicinal substances toward specific zones. The focus is here on the mechanical description of adhesion kinetics, in terms of the failure and creation of connections during the rolling phenomenon. Hence, we consider the case of a single cell, which is linked to a rigid substratum. A 2D model is established. We consider that the contact zone cell-wall is rectilinear and composed of vertical fibers and two horizontal rigid beams (complex cell membrane- fibers-vein wall). These connections are modeled by elastic springs having identical elastic properties (e.g stiffness), but different failure strengths. The cell is subjected to the flow of plasma, which can generate the rolling phenomenon; we accordingly consider two distinct zones, one associated with the failure of the old fibers and one with the creation of the new fibers (in the direction opposite to the flow). Several interactions are taken into account in this model: van der Waals (attractive) and electrostatic (repulsive) forces and the effects of fluid pressure, assimilated into a periodic point force applied to the interface zone. We also study the vibration induced failure in the contact zone without mechanical damping using the principle of virtual work and a failure criterion to establish the equation of motion and the time evolution of the failure (dynamical approach). Rupture of a fiber can occur if the stress applied to the fiber is above a certain limit. These limits are determinated with using a probabilistic approach by use of a spectral method to simulate a stochastic and Gaussian field. Modeling of the creation of new fibers is also achieved by the combination of a dynamical and probabilistic method and a kinematical criterion. On the basis of these elements, numerical simulations are developed, that elucidate the rupture and rolling phenomena.

Cell Adhesion↗
PubMed1988

A clinical diagnostic model for the assessment of asbestosis: a new algorithmic approach.

Asbestosis, one of the pneumoconioses that is defined by a set of clinical, radiographic, and pathologic findings, occurs as a result of exposure to asbestos fibers. Several approaches have attempted to describe the presence, progression, or extent of asbestosis. However, these approaches have attempted to describe the presence, progression, or extent of asbestosis. However, these approaches have limitations or lack correlations with other diagnostic modalities. We propose a comprehensive clinical diagnostic model that uses the sensitivities and specificities of the various clinical, radiographic, and pathologic findings to generate a set of "likelihood numbers." These likelihood numbers, contribute to the calculation of a value that can indicate the probability of asbestosis. The clinical diagnostic model is heuristic in that a specific feature supportive of the diagnosis of asbestosis may be tested as to its sensitivity and specificity, and new features may be added to the model. The model also indicates how probabilistic a given set of findings is in the diagnosis of asbestosis and suggests what additional data may make the diagnosis more or less statistically probable. Regarding the radiologic considerations of asbestosis, the strength of the clinical diagnostic model is that it is capable of supporting a diagnosis of asbestosis in the presence of a normal chest radiograph and, conversely, may reject the diagnosis of asbestosis despite the radiographic finding of pulmonary fibrosis.

Algorithms↗
PubMed1994

Fitting a mixture model by expectation maximization to discover motifs in biopolymers.

The algorithm described in this paper discovers one or more motifs in a collection of DNA or protein sequences by using the technique of expectation maximization to fit a two-component finite mixture model to the set of sequences. Multiple motifs are found by fitting a mixture model to the data, probabilistically erasing the occurrences of the motif thus found, and repeating the process to find successive motifs. The algorithm requires only a set of unaligned sequences and a number specifying the width of the motifs as input. It returns a model of each motif and a threshold which together can be used as a Bayes-optimal classifier for searching for occurrences of the motif in other databases. The algorithm estimates how many times each motif occurs in each sequence in the dataset and outputs an alignment of the occurrences of the motif. The algorithm is capable of discovering several different motifs with differing numbers of occurrences in a single dataset.

Algorithms↗