PubMed Health⌕ Search

SEARCH · PubMed Health

Results for “Model interpretability”

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 505 records · Page 28Linked to original sources

Dynamic selection of models for a ventilator-management advisor.

A ventilator-management advisor (VMA) is a computer program that monitors patients who are treated with a mechanical ventilator. A VMA implements a patient-specific physiologic model to interpret patient data and to predict the effects of alternative control settings for the ventilator. Because a VMA evaluates its physiologic model repeatedly during each cycle of data interpretation, highly complex models may require more computation time than is available in this time-critical application. On the other hand, less complex models may be inaccurate if they are unable to represent a patient's physiologic abnormalities. For each patient, a VMA should select a model that balances the tradeoff of prediction accuracy and computation-time complexity. I present a method to select models that are at an appropriate level of detail for time-constrained decision tasks. The method is based on a local search in a graph of models (GoM) for a model that maximizes the tradeoff of computation-time complexity and prediction accuracy. For each model under consideration, a belief network computes a probability of model adequacy given the qualitative prior information, and the goodness of fit of the model to the data provides a measure of the conditional probability of adequacy given the quantitative observations. I apply this method to the problem of model selection for a VMA. I describe an implementation of a graph of physiologic models that range in complexity from VentPlan, a simple model with 3 compartments, to VentSim, a multicompartment model with detailed airway, circulation and mechanical ventilator components.(ABSTRACT TRUNCATED AT 250 WORDS)

Computer Simulation↗

A case study on the choice, interpretation and checking of multilevel models for longitudinal binary outcomes.

Recent advances in statistical software have led to the rapid diffusion of new methods for modelling longitudinal data. Multilevel (also known as hierarchical or random effects) models for binary outcomes have generally been based on a logistic-normal specification, by analogy with earlier work for normally distributed data. The appropriate application and interpretation of these models remains somewhat unclear, especially when compared with the computationally more straightforward semiparametric or 'marginal' modelling (GEE) approaches. In this paper we pose two interrelated questions. First, what limits should be placed on the interpretation of the coefficients and inferences derived from random-effect models involving binary outcomes? Second, what diagnostic checks are appropriate for evaluating whether such random-effect models provide adequate fits to the data? We address these questions by means of an extended case study using data on adolescent smoking from a large cohort study. Bayesian estimation methods are used to fit a discrete-mixture alternative to the standard logistic-normal model, and posterior predictive checking is used to assess model fit. Surprising parallels in the parameter estimates from the logistic-normal and mixture models are described and used to question the interpretability of the so-called 'subject-specific' regression coefficients from the standard multilevel approach. Posterior predictive checks suggest a serious lack of fit of both multilevel models. The results do not provide final answers to the two questions posed, but we expect that lessons learned from the case study will provide general guidance for further investigation of these important issues.

Journal Article↗

Beyond the decoupling approximation in the model free approach for the interpretation of NMR relaxation of macromolecules in solution.

The model free (MF) approach allows straightforward extraction of generalized order parameters and correlation times for internal and overall bond vector reorientational fluctuations from NMR spin relaxation measurements for macromolecules in solution. The drawback of this approach is the use of a decoupling approximation that neglects correlations between internal and overall molecular motions. These correlations are significant when fluctuation amplitudes are less than the size of the "cage" that restricts the bond vector. In this regime, motion of the bond vector is only indirectly affected by molecule fluctuations through the orienting potential for the local director axis. By separating the regimes of large and small amplitude molecular fluctuations, an expression for the spectral density function is derived that does not invoke the decoupling approximation, but that relies on the same macroscopic level of description as the original MF approach. The parameters of the spectral density function provide insight into the effects on the MF parameters that result from violation of the decoupling approximation.

Computer Simulation↗

Development of linear, ensemble, and nonlinear models for the prediction and interpretation of the biological activity of a set of PDGFR inhibitors.

A QSAR modeling study has been done with a set of 79 piperazyinylquinazoline analogues which exhibit PDGFR inhibition. Linear regression and nonlinear computational neural network models were developed. The regression model was developed with a focus on interpretative ability using a PLS technique. However, it also exhibits a good predictive ability after outlier removal. The nonlinear CNN model had superior predictive ability compared to the linear model with a training set error of 0.22 log(IC50) units (R2 = 0.93) and a prediction set error of 0.32 log(IC50) units (R2 = 0.61). A random forest model was also developed to provide an alternate measure of descriptor importance. This approach ranks descriptors, and its results confirm the importance of specific descriptors as characterized by the PLS technique. In addition the neural network model contains the two most important descriptors indicated by the random forest model.

Linear Models↗

A new interpretation of ASM2d for modeling of SBR performance for enhanced biological phosphorus removal under different P/HAc ratios.

This study evaluated the prediction capability of Activated Sludge Model No. 2d (ASM2d), for the enhanced biological phosphorus removal (EBPR) performance of a sequencing batch reactor (SBR) receiving variable influent phosphate load. For this purpose, a laboratory-scale SBR was operated with a synthetic feed containing acetate as the sole carbon source. The experiments were conducted in four different Runs to ensure a range of different phosphate/acetate ratios in the influent. Model evaluations were carried out using concentration profiles measured throughout a representative cycle at steady state. An iterative calibration methodology was developed based on sensitivity analysis and applied to four different sets of experimental data on relevant model parameters reflecting SBR performance. ASM2d was able to predict the steady state behavior of the SBR system receiving variable influent phosphate loads only with the recalibrated parameter set. The regular changing pattern of the coefficients could be interpreted with the ability of the SBR system to sustain glycogen accumulating microorganisms, GAOs, which can store substrate under anaerobic conditions without polyphosphate energy, but deriving energy from the degradation of glycogen. Thus they are capable of prevailing at lower P/Ac ratios. The results indicate the need to include glycogen and GAOs as model components for processes involving both phosphate accumulating organisms, (PAOs) and GAOs, in order to obtain a better prediction of X(PHA) and oxygen uptake rate (OUR) profiles in the system.

Biodegradation, Environmental↗

GT-Mamba: a Topology-Aware Graph-State space model for robust and interpretable epigenetic age prediction.

MOTIVATION: Current epigenetic clocks face a trade-off between predictive accuracy and biological interpretability, often relying on dataset-specific correction to generalize across cohorts. We propose GT-Mamba, a novel architecture that integrates a Structure-Aware Graph Transformer with the Mamba state space model. This design captures CpG topological correlations and genome-wide long-range dependencies. RESULTS: GT-Mamba demonstrates strong out-of-the-box robustness across heterogeneous independent validation cohorts, achieving a weighted average MAE of 4.43 years. Notably, it effectively generalizes to EPIC 850k arrays despite partial feature missingness, and maintains consistent performance across homologous age distribution shifts (MAE 2.94 years in a young cohort). Ablation studies confirm that graph topology contributes to improved robustness against noise. Mechanistic analysis suggests that the model captures methylation patterns associated with both developmental and functional processes. AVAILABILITY: Source code and pre-trained models are freely available at https://github.com/NENUBioCompute/GT-Mamba and archived on Zenodo (DOI: 10.5281/zenodo.19703155).

Epigenesis, Genetic↗

Identification of antibody epitopes within the CB-11 peptide of type II collagen. II. Computer modelling studies of peptides and the interpretation of epitope scanning results.

Computer modelling techniques were used to investigate the structure of 8-mers from the CB-11 peptide of bovine type II collagen which were recognised by sera from rats which had previously been injected with bovine type II collage. It was discovered that all the hydrophobic peptides recognised by the rat sera were predicted to have collagenous-like secondary structures. The primary structure of the 8-mers which were recognised was also compared against the sequences in the OWL protein sequence database. The combined results of the computer modelling and sequence analysis suggested that the sequence Gly-Pro-Gly-Phe-Pro is a minimal B cell epitope of the CB-11 fragment of bovine type II collagen.

Amino Acid Sequence↗

A plausible molecular interpretation of the polar coordinate model with a centripetal degeneracy of distinctive field values.

I have attempted to interpret, in molecular terms, the polar coordinate model for epimorphic regulation, revised to accommodate the rigorously-held constraints of the polar coordinate system (Sibatani, 1981). If the angular coordinate of the field value is represented by a physical variable in the form of a proportion function, R, to be defined as the intensity ratio of two morphogen curves then R can stay unchanged along any radius mapped on the actual space, while the intensity of the two morphogens together approaches zero towards the field centre, satisfying the requirement of the centripetal degeneracy of the distinctive angular field values. Implications of this idea are discussed in theoretical and molecular terms as well as in respect of interpretations of some experimental data.

Animals↗

Validation of a model of left ventricular segmentation for interpretation of SPET myocardial perfusion images.

Several models of left ventricular segmentation have been developed that assume a standard coronary artery distribution, and are currently used for interpretation of single-photon emission tomography (SPET) myocardial perfusion imaging. This approach has the potential for incorrect assignment of myocardial segments to vascular territories, possibly over- or underestimating the number of vessels with significant coronary artery disease (CAD). We therefore sought to validate a 17-segment model of myocardial perfusion by comparing the predefined coronary territory assignment with the actual angiographically derived coronary distribution. We examined 135 patients who underwent both coronary angiography and stress SPET imaging within 30 days. Individualized coronary distribution was determined by review of the coronary angiograms and used to identify the coronary artery supplying each of the 17 myocardial segments of the model. The actual coronary distribution was used to assess the accuracy of the assumed coronary distribution of the model. The sensitivities and specificities of stress SPET for detection of CAD in individual coronary arteries and the classification regarding perceived number of diseased coronary arteries were also compared between the two coronary distributions (actual and assumed). The assumed coronary distribution corresponded to the actual coronary anatomy in all but one segment (#3). The majority of patients (80%) had 14 or more concordant segments. Sensitivities and specificities of stress SPET for detection of CAD in the coronary territories were similar, with the exception of the RCA territory, for which specificity for detection of CAD was better for the angiographically derived coronary artery distribution than for the model. There was 95% agreement between assumed and angiographically derived coronary distributions in classification to single- versus multi-vessel CAD. Reassignment of a single segment (segment #3) from the LCX to the LAD territory further improved the model's fit with the anatomic data. It is concluded that left ventricular segmentation using a model with assumed coronary artery distribution is valid for interpretation of SPET myocardial perfusion imaging.

Coronary Angiography↗

Electrical properties of frog skeletal muscle fibers interpreted with a mesh model of the tubular system.

This paper presents the construction, derivation, and test of a mesh model for the electrical properties of the transverse tubular system (T-system) in skeletal muscle. We model the irregular system of tubules as a random network of miniature transmission lines, using differential equations to describe the potential between the nodes and difference equations to describe the potential at the nodes. The solution to the equations can be accurately represented in several approximate forms with simple physical and graphical interpretations. All the parameters of the solution are specified by impedance and morphometric measurements. The effect of wide circumferential spacing between T-system openings is analyzed and the resulting restricted mesh model is shown to be approximated by a mesh with an access resistance. The continuous limit of the mesh model is shown to have the same form as the disk model of the T-system, but with a different expression for the tortuosity factor. The physical meaning of the tortuosity factor is examined, and a short derivation of the disk model is presented that gives results identical to the continuous limit of the mesh model. Both the mesh and restricted mesh models are compared with experimental data on the impedance of muscle fibers of the frog sartorius. The derived value for the resistivity of the lumen of the tubules is not too different from that of the bathing solution, the difference probably arising from the sensitivity of this value to errors in the morphometric measurements.

Animals↗

In silico model as a tool for interpretation of intestinal infection studies.

In nutrition research the number of human in vivo experiments is limited because of the many restrictions and the high costs of testing in humans. Up to now predictive computer models aiming to enhance research have been rare or too complex, with many nonmeasurable adjustable parameters. This study aimed to develop a basic physicochemical computer model for a first quantitative interpretation of results obtained from in vivo intestinal experiments with bacteria. This new modeling approach is validated with results obtained from gut infection studies in vivo. The design of the model is described, and its ability to reproduce experimental data is evaluated. The model predictions are compared with new experimental data. The phenomena that take place in the gastrointestinal tract are summarized by model constants for growth, adherence, and release of bacteria. Although the model is far from describing all details and many processes in the intestine are combined, the model calculation results lead to reasonable conclusions and interesting hypotheses. One of these hypotheses concluded from the model outcomes is that Escherichia coli bacteria have a much lower intestinal growth rate in humans than in rats. Extra laboratory validation experiments proved the reliability of this hypothesis predicted by the model. In addition, the known protective effect of dietary calcium and detrimental effect of clindamycin on the growth and adherence of Salmonella bacteria could be quantified. From these results it is clear that the model enhances the interpretation of in vivo gastrointestinal experiments and will facilitate research trajectories towards new functional foods that improve resistance to pathogenic bacteria in humans.

Animals↗

Mental models change rapidly with implicitly acquired information about the local environment: a two-tone image study.

The projection of 3-D objects to 2-D images necessitates a loss of information, thus the shape of volumetric objects depicted in images is inherently ambiguous. The results of 3 experiments suggest observers use mental models of the local visual environment to constrain image interpretation. These models change quickly and dramatically to accommodate implicitly acquired information. Observers viewed very high-contrast (2-tone) images of novel volumetric objects. Before priming, novel 2-tone images appeared 2-D. After incidental exposure to similar objects in grayscale or familiar objects in 2-tone, the test images appeared volumetric. Incidental leaming appears to alter observers' mental models, thus causing an alteration in image interpretation in the absence of any image change. Highlights were interpreted more accurately than shadows, suggesting shadows play a secondary role in shape recovery.

Adolescent↗

A mathematical high bar-human body model for analysing and interpreting mechanical-energetic processes on the high bar.

The aims of this study were: 1. To study the transfer of energy between the high bar and the gymnast. 2. To develop criteria from the utilisation of high bar elasticity and the utilisation of muscle capacity to assess the effectiveness of a movement solution. 3. To study the influence of varying segment movement upon release parameters. For these purposes a model of the human body attached to the high bar (high bar-human body model) was developed. The human body was modelled using a 15-segment body system. The joint-beam element method (superelement) was employed for modelling the high bar. A superelement consists of four rigid segments connected by joints (two Cardan joints and one rotational-translational joint) and springs (seven rotation springs and one tension-compression spring). The high bar was modelled using three superelements. The input data required for the high bar human body model were collected with video-kinematographic (50 Hz) and dynamometric (500 Hz) techniques. Masses and moments of inertia of the 15 segments were calculated using the data from the Zatsiorsky et al. (1984) model. There are two major phases characteristic of the giant swing prior to dismounts from the high bar. In the first phase the gymnast attempts to supply energy to the high bar-humanbody system through muscle activity and to store this energy in the high bar. The difference between the energy transferred to the high bar and the reduction in the total energy of the body could be adopted as a criterion for the utilisation of high bar elasticity. The energy previously transferred into the high bar is returned to the body during the second phase. An advantageous increase in total body energy at the end of the exercise could only be obtained through muscle energy supply. An index characterising the utilisation of muscle capacity was developed out of the difference between the increase in total body energy and the energy returned from the high bar. A delayed and initially slow but even reduction of hip and shoulder angles provided more advantageous release conditions. The total body energy could be improved by up to 15%, the vertical CM release velocity by up to 10% and the angular momentum by up to 35%.

Algorithms↗

Vestibular adaptation: how models can affect data interpretations.

Vestibular adaptation can be induced optically or by chemical or physical injury to the vestibular apparatus or the brain stem. In searching for the sites or mechanisms of vestibular adaptation, neurophysiologists often rely on comparing central resting (background) activities and central modulations (sensitivity) during vestibular stimulation, before and after motor learning or vestibular compensation. It is assumed that adapted central sites must exhibit modulation changes that parallel vestibulo-ocular reflex changes. Using model simulations and analysis, we will show that such presumptions may be misleading. First, using a simple schematic of interconnected cells or nuclei, one can show that modulation depth and background "tone" can be modified (or fixed) independently, using weightings on direct or indirect afferent projections. That is, if synaptic weights along all stimulus pathways are altered, one may fix or strongly modify central premotor characteristics in a manner apparently unrelated to global reflex changes. In the vestibulo-ocular reflex, the dominant premotor pathways contain position-vestibular-pause cells and eye-head-velocity cells (which are behaviorally similar to floccular-target neurons). Several experiments have reported negligible changes in the velocity sensitivity of position-vestibular-pause cells, despite large gain changes in the vestibulo-ocular reflex induced by training with visual-vestibular conflict. On the other hand, the modulation changes on floccular-target neurons (position-vestibular-pause) can be much larger than the changes in reflex gain. Using a bilateral vestibulo-ocular reflex model, we show that overall increases or decreases in reflex gain can be expressed (even overexpressed) in one particular subgroup of premotor neurons. Nevertheless, such observations are theoretically compatible with synaptic changes on all primary projections in a widely interconnected central network. Hence, stable neural responses during reflex adaptation are not sufficient to exclude a potential site of sensory-motor adaptation. Similarly, modified neural responses (as in cerebellum) need not necessarily imply a direct role in supporting the adapted state. Model predictions should help to design additional experimental protocols, to test hypotheses, and to refine diagnostic measures of recovery after vestibular lesions.

Adaptation, Physiological↗

An anatomical, histochemical, and autoradiographic study of the ever-growing molar dentition of Microtus with comments on the role of structure in growth and eruption.

An analysis of the microanatomy of the molar dentition of Microtus utilizing histological, histochemical and autoradiographic techniques reveals a complex architecture with distinctive morphogenic mechanisms which respond to the functional requirements of the organism. These mechanisms include; the maintenance of continued growth and eruption of the molars to compensate for continued hard tissue loss from wear at the occlusal surface of the crown throughout the entire lifespan of the orgainism and a positive feedback repair mechanism to protect the growth systems from the potential destruction this normal occlusal wear could initiate. An awareness and understanding of these phenomena is of significant value for interpreting palentological specimens and formulating a theoretical model for interpreting the evolution of Microtine molar dentitions.

Alveolar Process↗

Sorption Isotherms of Gases by Polymer Membranes in the Glassy State: An Explanation Based on the Nonequilibrium Thermodynamics.

Sorption isotherms of glassy polymers are concave to the pressure axis, and the absolute sorption levels are almost an order of magnitude higher than that of rubbery polymers on the relatively low-pressure side. There are several models to interpret this behavior, and the dual sorption model is the most widely accepted one among them. In the present work, sorption isotherms were first derived from permeation data with several gases for four polymer membranes which might be representative of the dual sorption model: two of them were in a glassy state (cellulose acetate and polyamide), the third one was a block copolymer which was composed of a glassy polymer and a rubbery polymer at the measuring temperature, and the fourth one was a Nafion membrane which was taken as a model membrane because of its channel structure with adsorption sites of charged groups supported by a rubbery polymer. These results did not necessarily support the dual sorption model. Subsequently, the validity of the underlying assumptions of two other sorption models for glassy polymers, that is, the matrix model and the deformation model, were examined and a new equation for the sorption isotherm with an ordering parameter was derived, which implied that the glassy polymer was in a nonequilibrium state and changed from the glassy to the rubbery state by absorbing the gas. Copyright 1998 Academic Press.

Journal Article↗

A quantitative contour analysis of axisymmetric vesicles spontaneously adhering onto a substrate.

The determination of membrane-substrate profile for adherent vesicle using confocal reflectance interference contrast microscopy (C-RICM) has pushed for the need of advanced mechanics model for interpreting adhesion mechanisms. In this work, a model for vesicles or cells adhesion is established, the governing equation is derived from the variation of the potential energy at the cohesive zone. A closed-form solution is found for vesicle spontaneously adheres to a substrate when its shear modulus, micro vanishes. Based on the model and C-RICM experiments the magnitude of the adhesion force is calculated for a lipid vesicle adheres to a glass substrate.

Biocompatible Materials↗

Longitudinal endolymph flow associated with acute volume increase in the guinea pig cochlea.

Endolymph volume disturbances were induced by microinjections of artificial endolymph into the second turn of the guinea pig cochlea at rates less than 60 nl/min. Induced longitudinal movements and area changes of endolymph were quantified in the basal turn using an ionic flow marker technique. Tetramethylammonium (TMA) was used as a flow marker by iontophoresing it into endolymph in micromolar amounts. TMA movements in the apical and basal directions were monitored by ion-selective electrodes. Changes in endolymph flow and cross-sectional area of scala media were derived using a mathematical model to interpret the recorded tracer time courses. The model was validated by performing comparable volume injections and flow measurements in fine-diameter plastic tubes. The rate of flow of endolymph measured prior to injection was close to zero, in agreement with prior studies. Based on the injection of different volumes into endolymph over a 15 min period, we found that injection of up to 80 nl of artificial endolymph into the second turn would not induce flow in the basal turn. However, above this amount, flow towards the base increased at a rate which correlated with the injected volume, with endolymph moving basally by a distance of 0.0067 mm/nl of artificial endolymph injected. Flow rates measured in the third turn, on the apical side of the injection were far lower and showed characteristics consistent with there being no outlet at the apex. These results suggest that small volume disturbances are corrected locally in the cochlea, but larger disturbances produce a longitudinal flow of endolymph out of the cochlea which represents a significant mechanism contributing to homeostasis. It can be concluded that structures outside the cochlea, such as the endolymphatic sac, do play a role in the correction of endolymph volume disturbances. Although the maintenance of endolymph composition is dominated by local ion transport mechanisms, the capacity of these local mechanisms to maintain normal endolymph volume appears to be limited.

Animals↗