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 181 records · Page 10Linked to original sources

PubMed1991

A framework for the interpretation of first-order interaction in logit modeling.

Several suggestions have been tendered for interpreting first-order interaction in log-linear analysis. Occasionally these methods result either in a loss of information or in results that are difficult to grasp on an intuitive level. It is argued that interpreting effect parameters in terms of odds ratios provides an elegant and intuitively appealing conceptual framework that has great generality across models. The interaction term then resembles a cross-product term in linear regression. In both forms of analysis, the partial effect of a given predictor on the response is composed of a constant and a correction that is a function of the other predictor involved in the interaction. This framework is especially appealing for models in which the logit is based on a bifurcation of the dependent variable. Although odds ratios are still useful for summarizing effects on polytomous dependent variables, greater caution must be exercised to avoid misleading interpretations.

Humans↗

PubMed1991

Combining physiologic models and symbolic methods to interpret time-varying patient data.

This paper describes a methodology for representing and using medical knowledge about temporal relationships to infer the presence of clinical events that evolve over time. The methodology consists of three steps: (1) the incorporation of patient observations into a generic physiologic model, (2) the conversion of model states and predictions into domain-specific temporal abstractions, and (3) the transformation of temporal abstractions into clinically meaningful descriptive text. The first step converts raw observations to underlying model concepts, the second step identifies temporal features of the fitted model that have clinical interest, and the third step replaces features represented by model parameters and predictions into concepts expressed in clinical language. We describe a program, called TOPAZ, that uses this three-step methodology. TOPAZ generates a narrative summary of the temporal events found in the electronic medical record of patients receiving cancer chemotherapy. A unique feature of TOPAZ is its use of numeric and symbolic techniques to perform different temporal reasoning tasks. Time is represented both as a continuous process and as a set of temporal intervals. These two temporal models differ in the temporal ontology they assume and in the temporal concepts they encode. Without multiple temporal models, this diversity of temporal knowledge could not be represented.

Adult↗

PubMed1975

Intrauterine pressure wave form characteristics in hypocontractile labor before and after oxytocin administration.

The data demonstrate that the contractions of hypocontractile active labor and normal spontaneous labor are different in several measures in addition to maximal amplitude. Furthermore, when the pathophysiology is corrected by the use of oxytocin, the contractions resemble those of normal spontaneous labor except in the maximal rate of tension development. Our data tend to support the subcellular model of uterine contractility, although the incompleteness of these models limits interpretation.

Adrenocorticotropic Hormone↗

PubMed2026

Prediction of antimicrobial minimum inhibitory concentration from bacterial genomes using a scalable and interpretable machine learning approach.

Although machine learning models can predict antimicrobial susceptibility from bacterial whole genome sequencing (WGS), state-of-the-art approaches are computationally demanding or dependent on knowledge of genetic resistance determinants. Here, we describe an efficient data-driven approach to predicting minimum inhibitory concentration (MIC) by progressively extending and refining predictive genome segments, independent of prior knowledge of resistance determinants. Resultant models had high interpretability - known and potentially novel resistance determinants were captured. Using 762 clinical E. coli strains, 71.6% of predictions were within one dilution of the measured MIC. Models trained with this algorithm generalised better onto external data (F1 score = 0.85) compared with alternative models trained on annotated resistance determinants (F1 = 0.82) or k-mer counts (F1 = 0.74). Computational demands were low (RAM usage 23.6GB vs 38.8GB for k-mer model). These advantages represent an important advance in predicting antimicrobial susceptibility from WGS, with potential applications for clinical diagnostics, drug development, and surveillance.

Journal Article↗

PubMed1991

Domain of validity of classical models of leucine metabolism assessed by compartmental modeling.

Whole-body modeling of in vivo leucine (an essential amino acid) metabolism is fundamentally difficult due to the complexity of the system. This has favored the use of two simple kinetic models, the so-called primary and reciprocal pool models, to interpret tracer data, but their domain of validity is uncertain. We define here the error of these two approaches by using comprehensive compartmental models of leucine metabolism as true representations of the leucine system. Of particular interest is the comparison of the two simple models with an 11-compartment model characterized by a rich intracellular compartmentation that has recently been proposed as a sound physiological description of the system. Formulas are derived that define in structural terms the error of the primary and reciprocal pool models.

Humans↗

PubMed1985

Model-independent electron spin resonance for measuring order of immobile components in a biological assembly.

A model-independent description of the angular orientation distribution of elements in an ordered biological assembly is applied to the electron spin resonance (ESR) technique. As in a previous model-independent treatment of fluorescence polarization (Burghardt, T.P., 1984, Biopolymers, 23:2383-2406) the elemental order is described by an angular distribution of molecular frames with one frame fixed in each element of the assembly. The distribution is expanded in a complete orthonormal set of functions. The coefficients of the series expansion (the order parameters) describe the orientation distribution of the elements in the assembly without reference to a model and can be obtained from the observed spectrum. The method establishes the limitations of ESR in detecting order in the assembly by determining which distribution coefficients the technique can detect. A method of determining the order parameters from an ESR spectra, using a set of ESR basis spectra, is developed. We also describe a treatment that incorporates the actual line shape measured from randomly oriented, immobile elements. In this treatment, no model-dependent assumptions about the line shape are required. We have applied the model-independent analysis to ESR spectra from spin-labeled myosin cross-bridges in muscle fibers. The results contain detailed information on the spin-probe angular distribution and differ in interesting ways from previous model-dependent interpretations of the spectra.

Animals↗

PubMed1987

[The results of research on the biological effects of neutrons].

A study was made of the peculiarities of the biological effect of neutrons as compared to gamma-radiation. A number of biological models were proposed for interpretation of the phenomenon: a biophysical model of mutations and cell inactivation, a model for interpretation of the formation of a radioresistant and radiosensitive fraction.

Animals↗

PubMed1989

Structures of actomyosin crossbridges in relaxed and rigor muscle fibers.

It was shown previously that a significant fraction of the myosin crossbridges is attached to actin in the skinned rabbit psoas fibers under relaxed conditions at low ionic strength and low temperature (Brenner, B., M. Schoenberg, J. M. Chalovich, L. E. Greene, and E. Eisenberg. 1982. Proc. Natl. Acad. Sci. USA. 79:7288-7291; Brenner, B., L. C. Lu, and R. J. Podolsky. 1984. Biophys. J. 46:299-306). In the present work, the structure of the attached crossbridges in the relaxed state between ionic strengths of 20 and 100 mM, as compared with that in the rigor state, is further examined by equatorial x-ray diffraction. Mass distributions projected along the fiber axis are reconstructed based on the first five equatorial reflections such that the spatial resolution is 128 A. The fraction of crossbridges attached under relaxed conditions are estimated to be in the range of 30% (at 100 mM ionic strength) and 60% (at 20 mM). The reconstructed density maps suggest that in the relaxed state, upon attachment the part of the crossbridge that centers around the thin filament is small, and the attachment does not significantly alter the center of mass of the myosin head distribution around the thick filament backbone. In contrast, accretion of mass in the rigor state occurs in a wider region surrounding the thin filament. In this case, mass in the surface region of the thick filament backbone is shifted slightly outward, probably by approximately 10 A. A schematic model for interpreting the present data is presented.

Actomyosin↗

PubMed1987

Evidence implicating alpha-2 adrenergic receptors in the anticonvulsant action of intranigral muscimol.

The effects of systemically administered catecholamine receptor antagonists on the anticonvulsant action produced by local application of a gamma-aminobutyric acid agonist, muscimol, to the substantia nigra of rats were studied. Both electroshock and kindled seizures were studied. Two alpha-2 receptor antagonists, idazoxan and yohimbine, blocked the anticonvulsant action of intranigral muscimol in the electroshock model. Neither a beta nor an alpha-1 adrenergic nor a dopamine-2 receptor antagonist blocked this action. In contrast to electroshock seizures, an alpha-2 antagonist only partially reversed the anticonvulsant action of intranigral muscimol in the kindling model. We interpret the data to indicate that the interaction of norepinephrine with alpha-2 receptors is required for the anticonvulsant action of intranigral muscimol in the electroshock model. The only partial reversal found in the kindling model suggests that nigral control of seizure propagation involves more than alpha-2 receptor mediated neurotransmission.

Animals↗

PubMed1992

A coupled pacemaker-slave model for the insect photoperiodic clock: interpretation of ovarian diapause data in Drosophila melanogaster.

A coupled circadian oscillator model for the insect photoperiodic clock is described which consists of a hierarchically arranged pacemaker and slave. The pacemaker is self-sustained, temperature compensated, and entrainable by the light cycle; the slave is a damping oscillation receiving entrainment from two sources, from the pacemaker via a coupling factor, and also directly from the light. The damping slave oscillation is seen as the "photoperiodic oscillator", equivalent to that proposed earlier by Lewis and Saunders (1987). The present simulations describe the effect of the strength of the coupling factor between hypothetical short- and long-period pacemaker oscillations (modelled on the "clock" mutants perS and perL2 in Drosophila melanogaster) and a slave oscillation with a period of about 24 hours. The output is presented in terms of photoperiodic response curves and Nanda-Hamner, or resonance, plots. With a high coupling strength, the pacemakers strongly entrain the slave, but with a low coupling strength the slave's properties are more evident. The model is presented as a possible explanation for recent ovarian diapause data in D. melanogaster "clock" mutants (Saunders 1990), but also as a more general model for the role of the insect circadian system in seasonal time measurement.

Animals↗

PubMed1991

Computer modelling of Tetrahymena axonemes at macromolecular resolution. Interpretation of electron micrographs.

A computer-generated model of the structural arrangement of the complete 9+2 ciliary axoneme of Tetrahymena at macromolecular resolution (4 nm) is presented. The model reconciles detailed information about subcomponents from negative-stained, thin-section and freeze-fracture electron micrographs, integrating the images into a consistent three-dimensional picture. This illuminates problems such as the requirement for compaction of dynein to form the arm, difficulties in visualization of the circumferential links, construction of the central sheath, and the comparative periodicities of the inner and outer arms. The model is pragmatic in that it is flexible and easily changed, as new information becomes available. It is also useful in the development of dynamic concepts, such as a spatial description of the dynein cross-bridge cycle, which is illustrated, or relationships between adjacent doublets during sliding and bending.

Animals↗

PubMed2026

Evaluating the pathogenic significance of unique chromosomal variants in craniosynostosis using patient-derived induced pluripotent stem cells and mouse modelling.

PURPOSE: Unravelling causal links between unique structural/copy-number variants (SV/CNV) and associated phenotypes is essential for correct genetic counselling. We investigated two families in which patients with craniosynostosis had SV/CNV potentially dysregulating a fibroblast growth factor (FGF)-encoding gene; a 730 kb dup(4)(q21.21) including FGF5; and a complex 568 kb interspersed 13q12.11 duplication, located 841 kb from FGF9. METHODS: We combined bioinformatic predictions of altered topologically-associating domain (TAD) structure, with experimental analysis (RNA- and ATAC- [assay for transposase-accessible chromatin] sequencing) of patient induced pluripotent stem cell lines (iPSCs) differentiated to neural crest (NCC) and osteoprogenitor (OPC) identities. For the dup(4)(q21.21) we generated a mouse bearing an equivalent rearrangement using CRISPR-Cas9 targeting. RESULTS: TAD analysis suggested potential dysregulation of the FGF5/FGF9 gene by bringing it into a novel genomic milieu. The RNA- and ATAC-seq assays demonstrated FGF5/FGF9 upregulation (2.7-18x) and local opening of chromatin, in 3/4 cell lines. For the dup(4)(q21.21), a causal role was supported by the mouse model, whereas interpretation of the 13q12.11 SV is confounded by a co-existing FOXP2 pathogenic variant. CONCLUSION: Patient iPSC-differentiated NCC and OPC lines, combined with TAD-based modelling to generate testable functional hypotheses, provide valuable functional evidence when evaluating causation of unique SV/CNV in craniosynostosis.

copy-number variant↗

PubMed2026

Causal circuit tracing reveals distinct computational architectures in single-cell foundation models: inhibitory dominance, biological coherence, and cross-model convergence.

MOTIVATION: Sparse autoencoders (SAEs) decompose foundation-model activations into interpretable features, but the model-internal causal interactions between those features (i.e. what ablating one feature does to the others, as distinct from the biological causal structure of the underlying cells)-and how those model-internal relationships relate to biological structure-are uncharacterized in single-cell foundation models. RESULTS: We introduce model-internal causal circuit tracing-zeroing one SAE feature at a source layer and measuring the resulting change in all downstream SAE features, for each of 120 source features-and apply it to Geneformer V2-316M and scGPT whole-human across four conditions (96 892 ablation-derived edges, 80 191 forward passes). On annotation-selected source features, edges share GO/KEGG/Reactome/STRING/TRRUST ontology terms at 50.9%-68.5%, a 2.9-6.2× enrichment over a configuration-preserving permutation null (P<.002); on 20 randomly sampled source features this attenuates to 21.5%-26.3%-still 2.5-3.1× above null-quantifying the annotation-selection contribution. Inhibitory dominance (fraction of ablation edges with d<0, i.e. source activation supports downstream target) is 65.5%-89.4%. scGPT produces larger raw per-edge effects (mean |d|=1.40 versus 1.05); after feature-share normalization, Geneformer is stronger (paired gene-pair ratio 0.64 on 33 301 shared pairs). Cross-model consensus yields 1142 architecture-invariant domain pairs (ordered pairs of GO biological-process categories "A→B" each connected by at least one ablation edge in both models; 10.6× enrichment over permutation null; P<.001). Circuit edge magnitude explains <1% of the variance in marginal driver-gene coexpression on the same cells (R2=0.010, n=31 176): the graph encodes structure beyond bivariate correlation. Against a matched-cell-type ENCODE ChIP-seq prior, circuit-predicted transcription factor (TF)→target pairs are enriched 2.06× (Fisher OR 5.84), markedly higher than 1.12× against TRRUST; direct ChIP-seq-supported target pairs show 10-30× larger CRISPRi sign-bias-corrected excess than indirect pairs. Gene-level CRISPRi validation on Replogle K562 and the noncancer RPE1 arm (and a true primary-T-cell control from Shifrut E, Carnevale J, Tobin V et al. Genome-wide CRISPR screens in primary human T cells reveal key regulators of immune function. Cell 2018; 175: 1958-71.e15) after sign-bias correction shows excess over baseline of +0.03 and +0.35 percentage points on K562 and RPE1, respectively (baseline already 52%-56% from sign marginals); effect-magnitude Spearman correlations ρ≈0. Bootstrap and per-cell-type stability (N∈{50,100,200}; B cell, CD4 + T, macrophage) give Pearson r≥0.97 on shared edges with 100% sign agreement; edge Jaccard grows monotonically with sample size. The circuit graph is therefore highly reproducible as an effect-size map, cell type specific in edge identity, consistent with coexpression encoding, and weakly but detectably enriched for ChIP-seq-supported direct regulatory edges. AVAILABILITY AND IMPLEMENTATION: https://github.com/Biodyn-AI/bio-sae-circuits (Python). Archival DOI: 10.5281/zenodo.19,633,166 (Zenodo).

Humans↗

PubMed1988

Porcine and human insulin absorption from subcutaneous tissues in normal and insulin-dependent diabetic subjects: a deconvolution-based approach.

The mechanisms of sc insulin absorption are not understood, and models for interpreting in vivo data cannot be developed without gross simplification. To overcome this difficulty we developed a new approach which makes use of deconvolution analysis and does not require any model of the sc tissue. In five normal subjects and seven insulin-dependent diabetic (IDDM) patients endogenous insulin secretion was suppressed by means of a hypoglycemic glucose clamp procedure (approximately 2.8 mmol/L) sustained by a continuous insulin infusion (approximately 4 pmol/min.kg). A bolus injection of insulin (5.4 nmol) was administered iv, and plasma insulin concentrations were measured frequently for 2 h to assess iv insulin kinetics. Insulin then was injected sc in the abdominal region, and plasma insulin concentrations were measured for 8 h. Each subject was studied twice, with porcine and semisynthetic human insulin (Actrapid, Novo). The rate of insulin absorption was reconstructed by deconvolution from the plasma concentrations and iv insulin kinetic data. Linearity of the iv insulin kinetics, essential for deconvolution analysis, was confirmed by a dose-response study in the range of the measured concentrations (150-1800 pmol/L). In most instances, a two-compartment model was adequate to describe the iv response. The mean plasma insulin clearance rates were 15.5 +/- 1.9 (+/- SD) mL/min.kg (porcine) and 17.2 +/- 6.0 (human) in normal subjects and 20.7 +/- 8.8 (porcine) and 20.9 +/- 9.1 (human) in the IDDM patients. The rate of appearance of human insulin from sc tissue was faster than that of porcine insulin in both normal and IDDM subjects, but no significant differences were found in bioavailability, which was 55 +/- 12% (+/- SD; porcine) and 61 +/- 34% (human) in the normal subjects, and 84 +/- 28% (porcine) and 86 +/- 23% (human) in the IDDM patients. The rate of absorption and bioavailability were higher in the IDDM patients than in the normal subjects, a difference possibly related to increased sc blood flow in the IDDM patients. No differences were found with regard to glucose requirement values, normalized to plasma insulin concentrations, in agreement with the finding that the bioavailability of the two insulin species was similar.

Adolescent↗

PubMed1991

Localization of the sites of pulmonary vasomotion by use of arterial and venous occlusion.

In this study, we present a new approach for using the pressure vs. time data obtained after various vascular occlusion maneuvers in pump-perfused lungs to gain insight into the longitudinal distribution of vascular resistance with respect to vascular compliance. Occlusion data were obtained from isolated dog lung lobes under normal control conditions, during hypoxia, and during histamine or serotonin infusion. The data used in the analysis include the slope of the arterial pressure curve and the zero time intercept of the extrapolated venous pressure curve after venous occlusion, the equilibrium pressure after simultaneous occlusion of both the arterial inflow and venous outflow, and the area bounded by equilibrium pressure and the arterial pressure curve after arterial occlusion. We analyzed these data by use of a compartmental model in which the vascular bed is represented by three parallel compliances separated by two series resistances, and each of the three compliances and the two resistances can be identified. To interpret the model parameters, we view the large arteries and veins as mainly compliance vessels and the small arteries and veins as mainly resistance vessels. The capillary bed is viewed as having a high compliance, and any capillary resistance is included in the two series resistances. With this view in mind, the results are consistent with the major response to serotonin infusion being constriction of large and small arteries (a decrease in arterial compliance and an increase in arterial resistance), the major response to histamine infusion being constriction of small and large veins (an increase in venous resistance and a decrease in venous compliance), and the major response to hypoxia being constriction of the small arteries (an increase in arterial resistance). The results suggest that this approach may have utility for evaluation of the sites of action of pulmonary vasomotor stimuli.

Animals↗

PubMed1975

Mathematical modelling of cell cycle and chronobiology: preliminary results.

A mathematical model taking into account the observed diurnal variations in cell kinetics is presented. The principle of the method is to divide each phase of the cell cycle in a definite number of compartments and to assume that the fluxes into and out of the compartments corresponding to the G1 phase are the only varying parameters through the day. Theoretical evolutions of percentages of cells in M and S phase, theoretical curves for percentage labelled mitosis experiment are derived. Preliminary results of the applications of the model to interpretation of published experimental data obtained in hamster cheek pouch epithelium are shown.

Cell Division↗

PubMed1975

[Psychoanalytic interpretation of sleep-disturbances. Model of a structural-theoretical classification (author's transl)].

In psychoanalytic literature anxiety, guilt-feelings and unconscious hostility are assumed to be the most common causes of spleeplessness. The author attempts to ascribe these emotions to conflicts between the instances of the psychoanalytic personality model as a structure-theoretical classification of sleep-disturbances. Three large groups emerge: Neurotic disturbances of sleep with internalized conflicts (these correspond essentially to what are generally understood to be neurotic sleep-disturbances), neurotic disturbances of sleep with externalized conflicts (these are more common in childhood) and non-neurotic sleep disturbances. Within these groups the relative parts of effectiveness of the instances of the psychoanalytic personality-model - ego, super-ego, id and reality - are discussed.

Aggression↗

PubMed1988

The electrical potential produced by a strand of cardiac muscle: a bidomain analysis.

Analytic expressions are derived relating the transmembrane potential to the intracellular, interstitial and external potentials in a cylindrical strand of cardiac muscle lying in a saline bath. The bidomain model is used to account for the anisotropy and interstitial space in the tissue. The implications of this model for interpreting potential data from strands of cardiac muscle are discussed.

Action Potentials↗