Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “statistical inference”

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 1,009 records · Page 56Linked to original sources
PubMed1993

Exploring the relationship between rationality and bounded rationality in medical knowledge-based systems.

If our goal in Artificial Intelligence in Medicine (AIM) is to engineer systems health-care providers will both use and, in the process, improve their performance, we must concentrate on the development of causal theories of knowledge and problem solving. One broad direction in pursuing this goal is understanding the relationships between existing models of rationality and bounded rationality for similar tasks. Models of rationality refer to those approaches in which the optimal properties of the models are deductively provable, i.e. in which the processing is rational. Representative models of rationality used in AIM are deductive logical models, statistical models such as Bayesian inference models, and decision-analytic models. Models of bounded rationality are those which do not guarantee such optimal properties nor yield to deductive correctness proofs. These models have their roots in cognitive psychology. In this article we show how explicating the relationship between models of rationality and bounded rationality might be done in the case of abductive tasks in medicine. This is done by positioning these modeling approaches within the same framework (an abstract computational model) and interpreting in this context both computational complexity results concerning the nature of the task and empirical results studies of human problem-solving behavior.

Artificial Intelligence↗
PubMed2006

Confounder selection in environmental epidemiology: assessment of health effects of prenatal mercury exposure.

PURPOSE: The purpose of the study is to compare different approaches to the identification of confounders needed for analyzing observational data. Whereas standard analysis usually is conducted as if the confounders were known a priori, selection uncertainty also must be taken into account. METHODS: Confounders were selected by using backward elimination (BE), change in estimate (CIE) method, Akaike information criterion, Bayesian information criterion (BIC), and an empirical approach using a priori information. A modified ridge regression estimator, which shrinks effects of confounders toward zero, also was considered. For each criterion, uncertainty in the estimated exposure effect was assessed by using bootstrap simulations for which confounders were selected in each sample. These methods were illustrated by using data for mercury neurotoxicity in Faroe Islands children. Point estimates and standard errors of mercury effects on confounder-sensitive neurobehavioral outcomes were calculated for each selection procedure. RESULTS: The full model and the empirical a priori model showed approximately the same precision, and these methods were (slightly) inferior to only modified ridge regression. Lower precisions were obtained by using BE with a low cutoff level, BIC, and CIE. CONCLUSIONS: Standard analysis ignores model selection uncertainty and is likely to yield overoptimistic inferences. Thus, the traditional BE procedure with p = 5% should be avoided. If data-dependent procedures are required for confounder identification, we recommend that inferences be based on bootstrap statistics to describe the selection process.

Animals↗
PubMed2004

Estimation of neurophysiological parameters from the waking EEG using a biophysical model of brain dynamics.

This paper presents the results from using electroencephalographic (EEG) data to estimate the values of key neurophysiological parameters using a detailed biophysical model of brain activity. The model incorporates spatial and temporal aspects of cortical function including axonal transmission delays, synapto-dendritic rates, range-dependent connectivities, excitatory and inhibitory neural populations, and intrathalamic, intracortical, corticocortical and corticothalamic pathways. Parameter estimates were obtained by fitting the model's theoretical spectrum to EEG spectra from each of 100 healthy human subjects. Statistical analysis was used to infer significant parameter variations occurring between eyes-closed and eyes-open states, and a correlation matrix was used to investigate links between the parameter variations and traditional measures of quantitative EEG (qEEG). Accurate fits to all experimental spectra were observed, and both inter-subject and between-state variability were accounted for by the variance in the fitted biophysical parameters, which were in turn consistent with known independent experimental and theoretical estimates. These values thus provide physiological information regarding the state. transitions (eyes-closed vs. eyes-open) and phenomena including cortical idling and alpha desynchronization. The parameters are also consistent with traditional qEEG, but are more informative, since they provide links to underlying physiological processes. To our knowledge, this is the first study where a detailed biophysical model of the brain is used to estimate neurophysiological parameters underlying the transitions in a broad range (0.25-50 Hz) of EEG spectra obtained from a large set of human data.

Awareness↗
PubMed2005

Applications of random field theory to electrophysiology.

The analysis of electrophysiological data often produces results that are continuous in one or more dimensions, e.g., time-frequency maps, peri-stimulus time histograms, and cross-correlation functions. Classical inferences made on the ensuing statistical maps must control family wise error (FWE) when searching across the map's dimensions. In this paper, we borrow multiple comparisons procedures, established in neuroimaging, and apply them to electrophysiological data. These procedures use random field theory (RFT) to adjust p-values from statistics that are functions of time and/or frequency. This RFT adjustment for continuous statistical processes plays the same role as a Bonnferonni adjustment in the context of discrete statistical tests. Here, by analysing the time-frequency decompositions of single channel EEG data we show that RFT adjustments can be used in the analysis of electrophysiological data and illustrate the advantages of this method over existing approaches.

Animals↗
PubMed1997

Older drivers and risk to other road users.

This paper examines the degree to which older drivers impose an "excess" risk of death or injury serious enough to require hospitalization on other road users; that is, the amount of risk older drivers impose on others above and beyond the amount imposed by drivers who are not yet old. A data set linking crash information from police accident reports to hospital discharge data in Wisconsin, 1991, was used to analyze excess risk associated with older drivers in two ways. First, the difference in the rate of serious injuries to other road users per 100 million driver age-group miles was used to estimate the total number of serious injuries resulting from the excess risk imposed on others by older drivers. Second, statistical models were used to infer the association between driver age and crash severity while conditioning on a variety of crash-specific information. Drivers aged 65-74 did not appear to impose excess risk of either deaths or injuries requiring hospitalization in either the aggregate or individual level analyses. Drivers aged 75 and over are associated with increased injuries to others, although the actual number is very small; the individual crash-level analysis suggests that a non-trivial part of the excess risk found in the aggregate analysis is a product of confounding.

Accidents, Traffic↗
PubMed2001

Estimating causal strength: the role of structural knowledge and processing effort.

The strength of causal relations typically must be inferred on the basis of statistical relations between observable events. This article focuses on the problem that there are multiple ways of extracting statistical information from a set of events. In causal structures involving a potential cause, an effect and a third related event, the assumed causal role of this third event crucially determines whether it is appropriate to control for this event when making causal assessments between the potential cause and the effect. Three experiments show that prior assumptions about the causal roles of the learning events affect the way contingencies are assessed with otherwise identical learning input. However, prior assumptions about causal roles is only one factor influencing contingency estimation. The experiments also demonstrate that processing effort affects the way statistical information is processed. These findings provide further evidence for the interaction between bottom-up and top-down influences in the acquisition of causal knowledge. They show that, apart from covariation information or knowledge about mechanisms, abstract assumptions about causal structures also may affect the learning process.

Adult↗
PubMed2003

Functional connectivity: studying nonlinear, delayed interactions between BOLD signals.

Correlation analysis has been widely used in the study of functional connectivity based on fMRI data. It assumes that the relevant information about the interactions of brain regions is reflected by a linear relationship between the values of two signals at the same time. However, this hypothesis has not been thoroughly investigated yet. In this work, we study in depth the information shared by BOLD signals of pairs of brain regions. In particular, we assess the amount of nonlinear and/or nonsynchronous interactions present in data. This is achieved by testing models reflecting linear, synchronous interactions against more general models, encompassing nonlinear, nonsynchronous interactions. Many factors influencing measured BOLD signals are critical for the study of connectivity, such as paradigm-induced BOLD responses, preprocessing, motion artifacts, and geometrical distortions. Interactions are also influenced by the proximity of brain regions. The influence of all these factors is taken into account and the nature of the interactions is studied using various experimental conditions such that the conclusions reached are robust with respect to variation of these factors. After defining nonlinear and/or nonsynchronous interaction models in the framework of general linear models, statistical tests are performed on different fMRI data sets to infer the nature of the interactions. Finally, a new connectivity metric is proposed which takes these inferences into account. We find that BOLD signal interactions are statistically more significant when taking into account the history of the distant signal, i.e., the signal from the interacting region, than when using a model of linear instantaneous interaction. Moreover, about 75% of the interactions are symmetric, as assessed with the proposed connectivity metric. The history-dependent part of the coupling between brain regions can explain a high percentage of the variance in the data sets studied. As these results are robust with respect to various confounding factors, this work suggests that models used to study the functional connectivity between brain areas should in general take the BOLD signal history into account.

Algorithms↗
PubMed1978

An analysis of the Co2+-induced nuclear magnetic resonance perturbations of hen egg white lysozyme.

A general methodology is presented for analyzing dipolar shifts induced by paramagnetic ions in the nuclear magnetic resonance (NMR) spectra of ligand molecules. The method is applied to the shift perturbations induced by Co2+ in the spectrum of hen egg white lysozyme. A hypothesis testing scheme is employed to evaluate statistically the relative precision with which the axially symmetric and non-axially symmetric forms of the dipolar shift equation fit the observed data. The assumption of axial symmetry for the magnetic susceptibility tensor of Co2+ is rejected at the confidence level of 99%. Since the results presented here are similar to those reached in our analysis of lanthanide-induced shifts, we suggest that the assumption of axial symmetry may, in general, not hold. Similar conclusions have been reached by other investigators in studies of paramagnetic metal binding to model systems. We have included the three Co2+ coordinates in an eight-parameter fit of the Co2+ shift data. The Co2+ position obtained from this fit is in statistical agreement with the position inferred from x-ray data. Thus, the analysis of shift data may furnish a means for determining the site of metal complexation in macromolecules whose structure has been determined by x-ray crystallography.

Animals↗
PubMed1986

Analyses of oviposition times and intervals in a wide range of layer flocks under normal and continuous lighting regimes.

Distributions of oviposition times and intra-clutch oviposition intervals in continuous light (CL) and normal light (NL) environments were studied in 39 populations of 27 different genotypes. The populations showed a range of mean intervals from 23.1 to 27.6 h in CL, arising from the history of selection directly on interval in some populations. The data were analysed to infer empirical relations of descriptive statistics of the distribution to each other and to rate of lay to 301 d of age (PRL). In NL, as the mean intra-clutch interval decreased, more intervals tended to accumulate against the barrier of 24 h imposed by the light-dark cycle and the variability of intervals declined rapidly. Similar, but less striking, relations were found in CL above a mean interval of about 24 h. Under CL, the mean intra-clutch interval (CIM) decreased beneath 24 h and the variability of intervals tended to show a steep increase. The rate of lay to 301 d of age increased linearly at a rate of 6.3% for each hour by which CIM decreased to about 24 h, but no further below that level. Oviposition time in NL was advanced with a decrease in CIM, apparently in two linear phases of differing slopes, perhaps reflecting different physiological bases of the change. The degree of entrainment in NL, measured by the proportion of eggs laid in the model 8 h, increased with reduction of CIM. The distribution of oviposition times in CL showed a great deal of variation among the populations and departed significantly (P less than 0.05) from the uniform rectangular distribution, in all but three populations. The proportion of eggs laid in the modal 8 h of the day, suggested as an indicator of sensitivity to uncontrolled timing cues in CL, was positively correlated with CIM. Free-running periods were estimated for individual pullets from sequences of oviposition times in CL. The mean period for a population declined linearly with CIM, but only when CIM had decreased below about 25.75 h, suggesting that the period of endogenous circadian rhythm might have been altered in some populations with low CIM's. The proportion of eggs laid in the modal 8 h of the free-running period, perhaps an indicator for regularity of the circadian rhythm, was negatively correlated with CIM.

Animals↗
PubMed2006

A comparison of phasing algorithms for trios and unrelated individuals.

Knowledge of haplotype phase is valuable for many analysis methods in the study of disease, population, and evolutionary genetics. Considerable research effort has been devoted to the development of statistical and computational methods that infer haplotype phase from genotype data. Although a substantial number of such methods have been developed, they have focused principally on inference from unrelated individuals, and comparisons between methods have been rather limited. Here, we describe the extension of five leading algorithms for phase inference for handling father-mother-child trios. We performed a comprehensive assessment of the methods applied to both trios and to unrelated individuals, with a focus on genomic-scale problems, using both simulated data and data from the HapMap project. The most accurate algorithm was PHASE (v2.1). For this method, the percentages of genotypes whose phase was incorrectly inferred were 0.12%, 0.05%, and 0.16% for trios from simulated data, HapMap Centre d'Etude du Polymorphisme Humain (CEPH) trios, and HapMap Yoruban trios, respectively, and 5.2% and 5.9% for unrelated individuals in simulated data and the HapMap CEPH data, respectively. The other methods considered in this work had comparable but slightly worse error rates. The error rates for trios are similar to the levels of genotyping error and missing data expected. We thus conclude that all the methods considered will provide highly accurate estimates of haplotypes when applied to trio data sets. Running times differ substantially between methods. Although it is one of the slowest methods, PHASE (v2.1) was used to infer haplotypes for the 1 million-SNP HapMap data set. Finally, we evaluated methods of estimating the value of r(2) between a pair of SNPs and concluded that all methods estimated r(2) well when the estimated value was >or=0.8.

Algorithms↗
PubMed1998

AMPA/kainate antagonist LY293558 reduces capsaicin-evoked hyperalgesia but not pain in normal skin in humans.

BACKGROUND: Animal studies suggest that alpha-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid-kainate (AMPA-KA) receptors are involved in pain processing. The effects of the competitive AMPA-KA antagonist LY293558 in two types of experimental pain in human volunteers, brief pain sensations in normal skin, and mechanical allodynia-pinprick hyperalgesia were studied after the injection of intradermal capsaicin. METHODS: Brief intravenous infusions of the competitive AMPA-KA antagonist LY293558 were given to 25 healthy volunteers to examine acute toxicity and analgesic effects. Fifteen volunteers then entered a double-blinded, three-period crossover study. In a Phase II study, LY293558 infusions (100% maximally tolerated dose vs. 33% maximally tolerated dose vs. placebo) began 10 min after intradermal injection of 250 microg capsaicin in volar forearm. Spontaneous pain, areas of mechanical allodynia and pinprick hyperalgesia, and side effects were determined every 5 min for 60 min. RESULTS: The median maximally tolerated dose was 1.3 +/- 0.4 (range, 0.9-2.0) mg/kg. Tests of cognitive and neurological function were unchanged. Dose-limiting side effects were hazy vision in 95% of volunteers and sedation in 40%. There were no significant changes in electrical or warm-cool detection and pain thresholds or heat pain thresholds. LY293558 had little effect on brief pain sensations in normal skin. Both high and low doses of LY293558 significantly reduced pain intensity, pain unpleasantness, and the area in which light brush evoked pain after intradermal capsaicin. There was a trend toward a dose-response effect of LY293558 on the area in which pinprick evoked pain after intradermal capsaicin, which did not reach statistical significance. CONCLUSIONS: The authors infer that AMPA-KA receptor blockade reduces the spinal neuron sensitization that mediates capsaicin-evoked pain and allodynia. The low incidence of side effects at effective doses of LY293558 suggests that this class of drugs may prove to be useful in clinical pain states.

Adult↗
PubMed1990

Meteorological modeling of radioiodine transport and deposition within the continental United States.

A meteorological transport and deposition interpolation model has been employed to estimate radioiodine deposition remote from the Nevada Test Site between points of observed deposition. The movement of the radioactive cloud across the United States is obtained from wind trajectories at various quasi-constant levels up to the top of the initial radioactive cloud. This model deals primarily with depositions beneath the passing cloud that are associated with precipitation. The fraction of radioiodine scavenged was inferred from a relationship derived statistically from fallout at locations having wet deposition. Therefore, estimation of deposition at locations between fallout observations is possible wherever precipitation occurred under the cloud. These estimates use very detailed historical records of precipitation available for virtually every county in the United States. Because significant dry and wet depositions also occur in areas that are not beneath the cloud, interpolation methods, such as kriging, are used to estimate fallout between the radioactivity observation locations. Ground-level radioactivity measurements are required for these alternative methods. Such observations are validated before being used, often by the use of supplementary low-level trajectories.

Computer Simulation↗
PubMed2004

Elucidation of gene interaction networks through time-lagged correlation analysis of transcriptional data.

The photosynthetic cyanobacterium Synechocystis sp. strain PCC 6803 uses a complex genetic program to control its physiological response to alternating light conditions. To study this regulatory program time-series experiments were conducted by exposing Synechocystis sp. to serial perturbations in light intensity. In each experiment whole-genome DNA microarrays were used to monitor gene transcription in 20-min intervals over 8- and 16-h periods. The data was analyzed using time-lagged correlation analysis, which identifies genetic interaction networks by constructing correlations between time-shifted transcription profiles with different levels of statistical confidence. These networks allow inference of putative cause-effect relationships among the organism's genes. Using light intensity as our initial input signal, we identified six groups of genes whose time-lagged profiles possessed significant correlation, or anti-correlation, with the light intensity. We expanded this network by using the average profile from each group of genes as a seed, and searching for other genes whose time-lagged profiles possessed significant correlation, or anti-correlation, with the group's average profile. The final network comprised 50 different groups containing 259 genes. Several of these gene groups possess known light-stimulated gene clusters, such as Synechocystis sp. photosystems I and II and carbon dioxide fixation pathways, while others represent novel findings in this work.

Cyanobacteria↗
PubMed2003

Dynamic force spectroscopy: optimized data analysis.

The forced rupture of single chemical bonds in biomolecular compounds (e.g., ligand-receptor systems) as observed in dynamic force spectroscopy experiments is addressed. An optimized method of data analysis is proposed. This method significantly outperforms the current standard one when applied to data from an idealized numerical computer simulation of an experiment with realistic parameter values. In particular, the force-free dissociation rate can be inferred with a considerably smaller statistical uncertainty and without the systematic overestimation of about 30%, which is shown to be inherent in the standard method.

Journal Article↗
PubMed1990

Comparison of dioptric power.

The important formal process of inferring properties of populations from statistics of samples requires a method by which variables can be compared or ranked. A suitable method of comparing dioptric powers is described. The method recognizes the essentially four-dimensional nature of dioptric power and is sufficiently general to allow for comparison of powers (for example, equivalent powers of general thick bitorics) that cannot be represented in the usual manner, sphere/cylinder x axis.

Eyeglasses↗
PubMed1978

Effect of atypical antibiotic resistance on microorganism identification by pattern recognition.

We classified microorganisms from the clinical laboratory by using information provided by the Gram stain and antibiotic sensitivity profiles obtained with the Bauer-Kirby technique. Approximately 4,000 microorganisms, routinely identified and tested for antibiotic sensitivities in a large hospital microbiology laboratory, were used as a data set for several pattern recognition classification methods: K--nearest-neighbor analysis, statistical isolinear multicomponent analysis, Bayesian inference, and linear discriminant analysis. K--nearest-neighbor analysis yielded the highest prospective classification accuracy for gram-negative organisms, 90%. When those organisms displaying an atypical antibiotic resistance pattern were excluded from the data, the gram-negative classification accuracy improved to 95%. These results are inferior to currently accepted biochemical identification methods. Microorganisms with atypical antibiotic resistance patterns are likely to be misidentified and are common enough (17% of our isolates) to limit the feasibility of routine identification of microorganisms from their antibiotic sensitivities.

Anti-Bacterial Agents↗
PubMed2000

Morphological identification of physiologically characterized afferents innervating the turtle posterior crista.

The turtle posterior crista consists of two hemicristae. Each hemicrista extends from the planum semilunatum to the nonsensory torus and includes a central zone (CZ) surrounded by a peripheral zone (PZ). Type I and type II hair cells are found in the CZ and are innervated by calyx, dimorphic and bouton afferents. Only type II hair cells and bouton fibers are found in the PZ. Units were intraaxonally labeled in a half-head preparation. Bouton (B) units could be near the planum (BP), near the torus (BT), or in midportions of a hemicrista, including the PZ and CZ. Discharge properties of B units vary with longitudinal position in a hemicrista but not with morphological features of their peripheral terminations. BP units are regularly discharging and have small gains and small phase leads re angular head velocity. BT units are irregular and have large gains and large phase leads. BM units have intermediate properties. Calyx (C) and dimorphic (D) units have similar discharge properties and were placed into a single calyx-bearing (CD) category. While having an irregular discharge resembling BT units, CD units have gains and phases similar to those of BM units. Rather than any single discharge property, it is the relation between discharge regularity and either gain or phase that makes CD units distinctive. Multivariate statistical formulas were developed to infer a unit's morphological class (B or CD) and longitudinal position solely from its discharge properties. To verify the use of the formulas, discharge properties were compared for units recorded intraaxonally or extracellularly in the half-head or extracellularly in intact animals. Most B units have background rates of 10-30 spikes/s. The CD category was separated into CD-high and CD-low units with background rates above or below 5 spikes/s, respectively. CD-low units have lower gains and phases and are located nearer the planum than CD-high units. In their response dynamics over a frequency range from 0.01-3 Hz, BP units conform to an overdamped torsion-pendulum model. Other units show departures from the model, including high-frequency gain increases and phase leads. The longitudinal gradient in the physiology of turtle B units resembles a similar gradient in the anamniote crista. In many respects, turtle CD units have discharge properties resembling those of calyx-bearing units in the mammalian central zone.

Algorithms↗
PubMed2000

Responses to efferent activation and excitatory response-intensity relations of turtle posterior-crista afferents.

Multivariate statistical formulas were used to infer the morphological type and longitudinal position of extracellularly recorded afferents. Efferent fibers were stimulated electrically in the nerve branch interconnecting the anterior and posterior VIIIth nerves. Responses of bouton (B) units depended on their inferred position: BP units (near the planum semilunatum) showed small excitatory responses; BT units (near the torus) were inhibited; BM units (in an intermediate position) had a mixed response, including an initial inhibition and a delayed excitation. Calyx-bearing (CD-high) units with an appreciable background discharge showed large per-train excitatory responses followed by smaller post-train responses that could outlast the shock train by 100 s. Excitatory responses were smaller in calyx-bearing (CD-low) units having little or no background activity than in CD-high units. Excitatory response-intensity functions, derived from the discharge during 2-s angular-velocity ramps varying in intensity, were fit by empirical functions that gave estimates of the maximal response (r(MAX)), a threshold velocity (v(T)), and the velocity producing a half-maximal response (v(1/2)). Linear gain is equal to r(MAX)/v(S), v(S) = v(1/2) - v(T). v(S) provides a measure of the velocity range over which the response is nearly linear. For B units, r(MAX) declines by as much as twofold over the 2-s ramp, whereas for CD units, r(MAX) increases by 15% during the same time period. At the end of the ramp, r(MAX) is on average twice as high in CD as in B units. Thresholds are negligible in most spontaneously active units, including almost all B and CD-high units. Silent CD-low units typically have thresholds of 10-100 deg/s. BT units have very high linear gains and v(S) < 10 deg/s. Linear gains are considerably lower in BP units and v(S) > 150 deg/s. CD-high units have intermediate gains and near 100 deg/s v(S) values. CD-low units have low gains and v(S) values ranging from 150 to more than 300 deg/s. The results suggest that BT units are designed to measure the small head movements involved in postural control, whereas BP and CD units are more appropriate for monitoring large volitional head movements. The former units are silenced by efferent activation, whereas the latter units are excited. This suggests that the efferent system switches the turtle posterior crista from a "postural" to a "volitional" mode.

Algorithms↗