SEARCH · PubMed Health

Results for “Model explainability”

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 631 records · Page 35Linked to original sources

PubMed2005

Predicting physiological concentrations of metabolites from their molecular structure.

Physiological concentrations of metabolites can partly be explained by their molecular structure. We hypothesize that substances containing certain chemical groups show increased or decreased concentration in cells. We consider here, as chemical groups, local atomic configurations, describing an atom, its bonds, and its direct neighbor atoms. To test our hypothesis, we fitted a linear statistical model that relates experimentally determined logarithmic concentrations to feature vectors containing count numbers of the chemical groups. In order to determine chemical groups that have a clear effect on the concentration, we use a regularized (lasso) regression. In a dataset on 41 substances of central metabolism in different organisms, we found that the physical concentrations are increased by the occurrence of amino and hydroxyl groups, while aldehydes, ketones, and phosphates show decreased concentrations. The model explains about 22% of the variance of the logarithmic mean concentrations.

Animals↗

PubMed2003

Detecting and measuring rater effects using many-facet Rasch measurement: part I.

The purpose of this two-part paper is to introduce researchers to the many-facet Rasch measurement (MFRM) approach for detecting and measuring rater effects. The researcher will learn how to use the Facets (Linacre, 2001) computer program to study five effects: leniency/severity, central tendency, randomness, halo, and differential leniency/severity. Part 1 of the paper provides critical background and context for studying MFRM. We present a catalog of rater effects, introducing effects that researchers have studied over the last three-quarters of a century in order to help readers gain a historical perspective on how those effects have been conceptualized. We define each effect and describe various ways the effect has been portrayed in the research literature. We then explain how researchers theorize that the effect impacts the quality of ratings, pinpoint various indices they have used to measure it, and describe various strategies that have been proposed to try to minimize its impact on the measurement of ratees. The second half of Part 1 provides conceptual and mathematical explanations of many-facet Rasch measurement, focusing on how researchers can use MFRM to study rater effects. First, we present the many-facet version of Andrich's (1978) rating scale model and identify questions about a rating operation that researchers can address using this model. We then introduce three hybrid MFRM models, explain the conceptual distinctions among them, describe how they differ from the rating scale model, and identify questions about a rating operation that researchers can address using these hybrid models.

Humans↗

PubMed1985

Uncertainty explains many aspects of visual contrast detection and discrimination.

More than 20 years ago, Tanner [Ann. N.Y. Acad. Sci. 89, 752 (1961)] noted that observers asked to detect a signal act as though they are uncertain about the physical characteristics of the signal to be detected. The popular assumptions of probability summation and decision variable, taken together, imply this uncertainty. This paper defines and uncertainty model of visual detection that assumes that the observer is uncertain among many signals and chooses the likeliest. With only four parameters, the uncertainty model explains why d' is approximately a power function of contrast ("nonlinear transduction") and accurately predicts effects of summation, facilitation, noise, subjective criterion, and task for near-threshold contrast. Thus the uncertainty model offers a synthesis of much of our current understanding of visual contrast detection and discrimination.

Attention↗

PubMed1996

Is referral source a risk factor for coronary surgery? Health maintenance organization versus fee-for-service system.

We began performing coronary artery bypass grafting for a large health maintenance organization (HMO) in 1974, as the sole provider of their cardiac surgery. The outcomes of our HMO group of patients were compared with those of our patients treated on a fee-for-service (FFS) basis. The HMO system entails preintervention and multidisciplinary screening conferences and is devoid of self-referral and personal financial incentives. Since 1985, the operative mortality for HMO patients has been consistently lower than for FFS patients. There were 8483 operations during this study period: 3168 (37%) were in the HMO group, with an overall operative mortality of 2.7%, and 5315 (63%) were in the FFS group, with an operative mortality of 4.6% (p=0.00002). This difference was investigated with univariate and multivariable analyses. Sixteen factors were found to univariately affect the risk of operative mortality; for five of these risk correlates there was a significant maldistribution between the HMO and FFS patients. Logistic regression was used to explore the influence of this imbalance in risk factors. The model found seven independent risk factors (left ventricular failure, emergency coronary bypass, redo bypass, nonuse of the internal thoracic artery, unstable angina, age, and diabetes) that significantly affected operative mortality. The FFS group variable closely approached independent risk significance at p=0.059. This multivariable model explained only one third of the observed differences in actual mortality between the HMO and FFS groups. The system-wide angioplasty/coronary bypass ratio, which could not be used in a patient-specific model, was 0.6 in the HMO system and 1.5 in the FFS group. Other factors related to the operating structure of a mature, large HMO may account for the remainder of the difference. The HMO referral system, through a powerful selection process, resulted in fewer emergencies, redo bypass operations, and catheterization complications that, in turn, yielded lower operative mortality than a noncoordinated FFS system of cardiovascular management.

Angioplasty, Balloon, Coronary↗

PubMed1997

Parental distress and children's problems among single-parent families in China.

The coercion model explains reciprocal relationships between parents' and children's adjustment problems, with the mediation of parenting behavior and social relationships. A survey of 301 single parents in Guangzhou, China, was performed to test such a model with reference to parental distress, perceived behavioral problems and anxiety of the eldest child, acceptance of the child, and experienced social pressure and social support. Structural equation modeling demonstrated a good fit of the coercion model as a theoretically based and simplified representation of the relationships, supporting the hypotheses that (a) the parent's acceptance helps prevent the child's behavioral problems, (b) the child's behavioral problems and anxiety contribute to the parent's distress, (c) the child's behavioral problems and anxiety invite social pressure on the parent, (d) social pressure on the parent aggravates and social support for the parent attenuates the parent's distress, (e) social pressure on and social support for the parent facilitate the parent's acceptance of his or her child. However, the effect of parental distress on acceptance of the child was not significant.

Anxiety↗

PubMedDate not supplied

Women's views of cardiac rehabilitation programs.

BACKGROUND: Previous approaches to understanding participation in cardiac rehabilitation programs based on middle-aged men may not be useful to explain the participation of women. The focus group research method produces unique data about attitudes and experiences that can serve as an adjunct to other forms of data collection in an effort to develop models explaining cardiac rehabilitation participation. The aim of this pilot study was to examine the perceptions and experiences of women participating in a cardiac rehabilitation program using focus groups. METHODS: Ten women (mean age 72 years) who had recently participated in a cardiac rehabilitation program were interviewed in focus groups of 6 and 4 members each using a semi-structured interview guide. Verbatim transcriptions of the focus group interviews were analyzed by two individuals using content analysis methods. RESULTS: Results indicated that prior to participation women had no idea what to expect in a cardiac rehabilitation program. Features women liked most about the program included: (1) feeling "safe" during exercise because they were monitored; (2) peer group support during rehabilitation; and (3) pleasant, encouraging staff. Women desired more: (1) social interaction during the cardiac rehabilitation exercise sessions; (2) emotional support from staff members about all dimensions of cardiac recovery; and (3) exercise options other than cycle or treadmill. CONCLUSIONS: Several cardiac rehabilitation program design and operational features are perceived by women as not meeting their needs. Although limited by its small sample size, findings from this study provide valuable insights into women's perceptions of cardiac rehabilitation programs.

Aged↗

PubMed1987

Transplacental clearance and blood flows of bovine gravid uterus at several stages of gestation.

Rates of uterine and umbilical blood flows and transplacental clearance of deuterium oxide (D2O) were determined for cows on 137 +/- 1.0 (SE, n = 9), 180 +/- 0.5 (n = 8), 226 +/- 0.4 (n = 9), and 250 +/- 1.8 (n = 5) days of gestation. From days 137 to 250, rates increased 4.5-fold for uterine blood flow, 21-fold for umbilical blood flow, and 14-fold for clearance of D2O. Changes in rates of umbilical blood flow and D2O clearance paralleled increased rates of fetal growth and metabolism, which have previously been reported to occur during the last half of gestation. The regressions of D2O clearance on uterine and umbilical blood flows were significant (P less than 0.01) and explained 94-99% of the variation in placental clearance of D2O. Because the rate of D2O clearance was always less than that of uterine and umbilical blood flows, and because a relatively simple statistical model explained most of the variation in clearance, it was suggested that a concurrent or countercurrent arrangement of maternal and fetal placental microvasculatures is not adequate to explain clearance of highly diffusable substances across the bovine placenta. In addition, a placental exchange diagram of the data showed the existence of severe uneven distribution of maternal and fetal placental blood flows and/or significant shunting of maternal and fetal placental flows away from areas of exchange. Taken together, these data indicate that the placenta of the cow, like those of the sheep and goat, represents a relatively inefficient system of transplacental exchange.

Animals↗

PubMed2004

Picolinic acids as inhibitors of dopamine beta-monooxygenase: QSAR and putative binding site.

Dopamine beta-monooxygenase (DBM, EC 1.14.17.1) catalyzes the oxidation of dopamine into (R)-noradrenaline. DBM inhibitors may act as antihypertensive drugs. A series of 22 picolinic acids substituted in 4- and 5-position was previously synthesized and tested for inhibition of DBM from bovine adrenal medulla. The QSAR of these compounds were investigated by Hansch analysis and comparative molecular field analysis (CoMFA). The correlation of pI(50) values with electronic (nucleophilic substituent constant sigma(p) (-), oxygen net charges and highest occupied molecular orbital energy calculated by AMPAC-AM1), hydrophobic (pi values of R(4)) and steric descriptors (molar refraction and Sterimol parameters of R(5)) indicated that a more negatively charged carboxylate moiety, more lipophilic R(4) groups as well as wider bulk and higher molar refraction of 5-substituents increase DBM inhibition. The CoMFA approach generally reproduced these QSAR in terms of steric and electrostatic field variables, the latter restricted to the carboxylate area. To predict a putative binding site, dopamine and fusaric acid were docked into a partial homology model of DBM derived from a crystal structure of peptidylglycine alpha-hydroxylating monooxygenase (EC 1.14.17.3). The inhibitor is suggested to interact by its carboxylate group with the copper site CuB and the protonated amino group of dopamine according to the uncompetitive type of inhibition. R(4) points to a tyrosine side chain. R(5) protrudes into the fringe of the catalytic crevice. It may "freeze" to the solvated surface of polar amino acids and additionally contact an isoleucine residue. Taken together, the model explains the QSAR results by corresponding types of interaction.

Adrenal Medulla↗

PubMed2000

Scale-invariant fluctuations at different levels of organization in developing heart cell networks.

In the last few years it has been realized that the intervals of spontaneous spiking events in the intact heart exhibit coexisting scale-invariant count fluctuations and anticorrelated interspike intervals fluctuations. Here, we show experimentally that this feature is an intrinsic property of single isolated heart cells, which is preserved when the cells couple into networks. We present a model explaining this behavior at both the single cell and network levels.

Animals↗

PubMed2003

Simultaneous effects of environmental factors on motile Aeromonas dynamics in an urban effluent and in the natural seawater.

Seasonal dynamics of motile Aeromonas in a treated urban effluent and in natural seawater along the Sfax coast (Mediterranean sea, Tunisia) were measured over a year concurrently with seven environmental factors, and compared with those of faecal coliforms. Counts for Aeromonas from a standard plate count method, ranged from 1.48 x 10(5)CFU.100 ml(-1) to 2.2 x 10(8)CFU.100 ml(-1) in the effluent and from 7.9 x 10(3)CFU.100 ml(-1) to undetectable level in the surface marine waters. Contrary to faecal coliforms, the Aeromonas dynamics exhibited a seasonal distribution in seawater which was inverse of the seasonal distribution in the sewage: From the end of November 1998 to April 1999 (cold period), Aeromonas counts increased in the treated effluent, while it decreased very rapidly in seawater. From May to October (warm period), Aeromonas abundance decreased in the effluent but showed an increasing fluctuating trend in the marine waters with a maximum in late summer/early autumn when the temperatures were around 22-23 degrees C. Multiple correlation and regression analyses suggest, by the coefficient of determination (R(2)), that 42% of variance in Aeromonas number changes in the treated effluent, may be explained by only turbidity, radiation and Aeromonas density in the previous sample, while 37% of variance in marine ecosystem were explained by radiance and conductivity. Furthermore, the t statistics and their p values and the coefficient of partial determination (r(2)) indicated that radiance contributed the most (r(2)=0.3184, t=-3.2, p=0.0041) to the dynamics of motile Aeromonas in seawater, when combined with conductivity. The models relevant for changes in faecal coliforms abundance incorporated turbidity, radiance in the effluent and conductivity, pH, radiance, turbidity in coastal marine environment. These models explain 66% and 73% of the observed cell number fluctuation, with turbidity (r(2)=0.529, t=5.08, p=0.0001) and conductivity (r(2)=0.5407, t=4.97, p=0.0001) as dominant factors in the multivariate model proposed, respectively, for the two sampling sites. The results presented here suggest that the combination of negative effects of sunlight and conductivity in natural seawater mainly affects the colony-forming capacity and make the motile Aeromonas nonrecoverable during cold months.

Aeromonas↗

PubMed2006

Contribution of cognitive factors to the prediction of post-traumatic stress disorder, phobia and depression after motor vehicle accidents.

Past research into the psychological consequences of traumatic events has largely focused on post-traumatic stress disorder (PTSD), although other anxiety disorders and depression are also common in the aftermath of trauma. Little is known about differential predictors of these conditions. The present study investigated the extent to which theoretically derived cognitive variables predict PTSD, phobias and depression after motor vehicle accidents. The cognitive predictors were compared to a set of established, mainly non-cognitive predictors. In addition, we tested how disorder-specific the cognitive predictors are. Participants (n=101) were interviewed within a year after having been injured in a motor vehicle accident. Diagnoses of PTSD, travel phobias and depression, symptom severities and predictor variables were assessed with self-report questionnaires and structured interviews. In multiple regression analyses, the sets of cognitive variables derived from disorder-specific models explained significantly greater proportions of the variance of the symptom severities than the established predictors (PTSD 76% vs. 45%, depression 72% vs. 46% and phobia 66% vs. 40%), and than cognitive variables derived from the models of the other disorders. In addition, the majority of individual cognitive variables showed the expected pattern of differences between diagnostic groups. The results support the hypothesis that disorder-specific sets of cognitive factors contribute to the development and maintenance of PTSD, phobias and depression following traumatic events.

Accidents, Traffic↗

PubMed2005

Do cortical neurons process luminance or contrast to encode surface properties?

On the one hand, contrast signals provide information about surface properties, such as reflectance, and patchy illumination conditions, such as shadows. On the other hand, processing of luminance signals may provide information about global light levels, such as the difference between sunny and cloudy days. We devised models of contrast and luminance processing, using principles of logarithmic signal coding and half-wave rectification. We fit each model to individual response profiles obtained from 67 surface-responsive macaque V1 neurons in a center-surround paradigm similar to those used in human psychophysical studies. The most general forms of the luminance and contrast models explained, on average, 73 and 87% of the response variance over the sample population, respectively. We used a statistical technique, known as Akaike's information criterion, to quantify goodness of fit relative to number of model parameters, giving the relative probability of each model being correct. Luminance models, having fewer parameters than contrast models, performed substantially better in the vast majority of neurons, whereas contrast models performed similarly well in only a small minority of neurons. These results suggest that the processing of local and mean scene luminance predominates over contrast integration in surface-responsive neurons of the primary visual cortex. The sluggish dynamics of luminance-related cortical activity may provide a neural basis for the recent psychophysical demonstration that luminance information dominates brightness perception at low temporal frequencies.

Animals↗

PubMed2005

Elastic lever-arm model for myosin V.

We present a mechanochemical model for myosin V, a two-headed processive motor protein. We derive the properties of a dimer from those of an individual head, which we model both with a four-state cycle (detached; attached with ADP.Pi; attached with ADP; and attached without nucleotide) and alternatively with a five-state cycle (where the powerstroke is not tightly coupled to the phosphate release). In each state the lever arm leaves the head at a different, but fixed, angle. The lever arm itself is described as an elastic rod. The chemical cycles of both heads are coordinated exclusively by the mechanical connection between the two lever arms. The model explains head coordination by showing that the lead head only binds to actin after the powerstroke in the trail head and that it only undergoes its powerstroke after the trail head unbinds from actin. Both models (four- and five-state) reproduce the observed hand-over-hand motion and fit the measured force-velocity relations. The main difference between the two models concerns the load dependence of the run length, which is much weaker in the five-state model. We show how systematic processivity measurement under varying conditions could be used to distinguish between both models and to determine the kinetic parameters.

Adenosine Diphosphate↗

PubMed2000

A structural model for predicting mathematics achievement: its relation with anxiety and self-concept in mathematics.

This study proposed and tested a model of mathematics achievement and its relations to antecedent and subsequent factors using structural equations modeling. A sample of elementary school students in Al-Ain school district (n = 394) completed an Arabic version of the Self-description Questionnaire as well as a questionnaire measuring their perception of the importance of mathematics, anxiety about it, and the amount of effort they exerted in studying. Mathematics grades were obtained from the official school records. Importance and effort were positively related to achievement which in turn had a positive path coefficient to self-concept and a negative path to anxiety. The hypothesized model explained 40%, 64%, and 73% of the variance in achievement, self-concept, and anxiety, respectively. The results can be interpreted as indicating that achievement is an important outcome and antecedent construct within the proposed model.

Achievement↗

PubMed2003

A spatial analysis of county-level variation in hospitalization rates for low back problems in North Carolina.

Hospitalization rates for low back problems vary widely. In previous non-spatial analyses, population-level socioeconomic and health resource characteristics have explained little of the variation in rates. This study examines geographic variation in hospitalization rates for low back problems while controlling for spatial dependence in the data. County-level surgical and medical hospitalization rates were calculated using North Carolina hospital (USA) discharge data from 1990-92. Non-spatial and spatial regression models were estimated using socioeconomic and health resource predictors. Both surgical and medical rates varied significantly among the 100 counties. Non-spatial models explained 62% of variation in log-transformed surgical rates and 66% of variation in log-transformed medical rates; however, residuals showed significant spatial dependence. Spatial lag models were therefore applied. Using simple contiguity spatial weights, surgery rates increased with higher percent urban population, primary care physician density, and discharge rate for other causes, and decreased with higher percent college graduates, percent disabled, occupied hospital bed density, and unoccupied hospital bed density. There was a nonlinear relationship between surgery rates and percent employed in heavy lifting/transportation industries. Medical rates increased with higher other-cause discharge rate and with MRI/CT scanner availability, and decreased with higher percent urban population, percent nonwhite population, percent in heavy lifting/transportation industries, and unoccupied hospital bed density. The results show that population-level socioeconomic and health resource characteristics are important determinants of variation in low back hospitalization rates. Independent of these variables, a separate spatial process produces geographic clustering of high-rate counties. Spatial effects are important and should be considered in small area analyses.

Catchment Area, Health↗

PubMed1993

Modeling lag phase of nonproteolytic Clostridium botulinum toxigenesis in cooked turkey and chicken breast as affected by temperature, sodium lactate, sodium chloride and spore inoculum.

The length of the lag phase (LP) of toxigenesis in commercially cooked turkey meat stored under vacuum was determined as affected by 0, 1.2, 2 and 3% sodium lactate (L), 0, 1 and 2% NaCl (S), spore (pool of nonproteolytic B and E strains: B2, B17, B197, B706, E211, E250, E KA-2 and E Beluga) inoculum (I) of 10(-2) to 10(4), storage temperature (T) of 4, 8, 12, 16, 20 and 30 degrees C and their interactions. The time from inoculation to the detection of first toxic sample was defined as LP. Using regression analysis the following model predictive of LP of C. botulinum toxigenesis in the cooked turkey breast was derived: Log(1/LP) = -2.2877 -0.1235(S) -0.2174(L) +0.4391(square root of T) +0.0204(square root of T) (I). The model explained 94.5% of the variation in results, in which square root of T was the most influential factor (65%), followed by L (21.2%), interaction of I and square root of T (4.9%) and S (3.4%). The model predicted LPs longer than those observed in 28.3% of the comparisons, but only in 1% of the comparisons when the lower limit of the 90% confidence interval of LP was used. Similar trends on the effect of L on C. botulinum were observed in an inoculated chicken meat study. This study demonstrated quantitatively that increasing L and S concentrations and lowering of T had a beneficial effect on delaying toxigenesis.

Analysis of Variance↗

PubMed2007

The simultaneous type, serial token model of temporal attention and working memory.

A detailed description of the simultaneous type, serial token (ST2) model is presented. ST2 is a model of temporal attention and working memory that encapsulates 5 principles: (a) M. M. Chun and M. C. Potter's (1995) 2-stage model, (b) a Stage 1 salience filter, (c) N. G. Kanwisher's (1987, 1991) types-tokens distinction, (d) a transient attentional enhancement, and (e) a mechanism for associating types with tokens called the binding pool. The authors instantiate this theoretical position in a connectionist implementation, called neural-ST2, which they illustrate by modeling temporal attention results focused on the attentional blink (AB). They demonstrate that the ST2 model explains a spectrum of AB findings. Furthermore, they highlight a number of new temporal attention predictions arising from the ST2 theory, which are tested in a series of behavioral experiments. Finally, the authors review major AB models and theories and compare them with ST2.

Attention↗

PubMedDate not supplied

The German Environmental Survey 1990/92 (GerES II): primary predictors of blood cadmium levels in adults.

As part of the representative Environmental Survey in Germany in 1990-1992, investigators determined cadmium levels in blood provided by 3,965 subjects aged 25-69 yr. The investigators considered approximately 150 variables (i.e., demographics, household and occupational characteristics, environmental exposures, smoking habits, frequency of food consumption, and additional life-style features) as potential predictors in multivariate regression analysis. On the basis of the results of multivariate regression analysis, the authors derived 2 slightly different models for the prediction of blood cadmium levels in populations from West and East Germany. Both models included 3 primary predictors of blood cadmium levels. The 2 models explained 51.3% and 61.2% of the observed variance in blood cadmium levels in West and East Germany, respectively. The most important predictor was a specific indicator for smoking habits, which was determined from a separate mathematical model. In this model, the effect of smoking was considered, and the model accounted for the biological half-life of cadmium in blood.

Adult↗