Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Linear Regression”

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 1,153 records · Page 64Linked to original sources
PubMed2005

Smoke-free law did not affect revenue from gaming in Delaware.

OBJECTIVE: To determine the effect of the Delaware smoke-free law on gaming revenue. METHODS: Linear regression of gaming revenue and average revenue per machine on a public policy variable, time, while controlling for economic activity and seasonal effects. RESULTS: The linear regression showed that the smoke-free law was associated with no effect on total revenue or average revenue per machine. CONCLUSION: Smoke-free laws are associated with no change in gaming revenue.

Delaware↗
PubMed2005

A robust algorithm for ratio estimation in two-color microarray experiments.

The reliability of the algorithms for ratio estimation in two-color microarray image analysis is very important, as these ratios build up the primary source of information for the subsequent analytical procedures (normalization, clustering, classification, etc). Although various algorithms already exist, there is still a need to develop procedures having higher levels of accuracy and robustness. We present a statistical procedure for the detection and removal of aberrant pixels in two-color microarray images. It is based on a linear regression approach, assuming reasonably high level of correlation between the two color channels. This procedure ensures more robust ratio estimation for the spots in both linear regression and traditional segmentation algorithms. The developed algorithms have been evaluated using simulated artificial images and experimental images of different designs. A demonstration version of the software can be downloaded from http://bioinfo.curie.fr/projects/maia/.

Algorithms↗
PubMed1998

Which arterial and cardiac parameters best predict left ventricular mass?

BACKGROUND: Many cardiovascular and noncardiovascular parameters are thought to be determinants of left ventricular mass (LVM). Complicated interactions necessitate the simultaneous measurement and consideration of each to determine their individual and collective impact on LVM. We undertook such a comprehensive study. METHODS AND RESULTS: The influence of anthropometry, cardiac size and contractility, arterial structure and function, as well as indices of lifestyle, physical activity, and dietary salt intake on LVM (by two-dimensionally guided M-mode echocardiography) was analyzed in 1315 Chinese subjects who were either normotensive or had untreated hypertension. Effects of many cardiac and arterial factors were assessed. In univariate analysis, almost all measured noncardiovascular, cardiac, and arterial variables were significantly correlated with LVM. In multivariate linear regression analyses, when age, sex, body habitus, fasting serum C-peptide level, dietary salt, physical activity, and lifestyle were accounted for, the optimum multivariate linear regression main effects model had an adjusted model r2 of 0.740, with 98% of the model variance accounted for by the 5 independent determinants of LVM: stroke volume (49.6%), systolic blood pressure (30.7%), contractility (14.7%), body mass index (1.8%), and aortic root diameter (1.6%). Other proposed arterial indices were significant independent determinants of LVM only when blood pressure was removed from the model and, even then, these indices not only resulted in less powerful prediction but also accounted for only a very small percentage of the total variance of LVM. CONCLUSIONS: In a large population, we (1) confirmed that age, body habitus, and some indexes of arterial structure and function are independent determinants of LVM; (2) found aortic diameter to be an independent structural determinant of LVM; (3) demonstrated that the effects of the derived measures of arterial function were small and provided no better predictive power than blood pressure alone; and (4) showed that when the best measures of cardiac and vascular load were included, the single most potent predictor was an index of left ventricular size.

Adult↗
PubMed1996

Statistical modelling reveals demography and time are the main contributing factors in global sperm count changes between 1938 and 1996.

Declining sperm count reports have caused enormous concern to both the scientific community and to society. We reproduced the linear regression analysis and the quadratic model analysis using the 50 year sperm count data published in Carlsen's report and found that neither model adequately described the data. The reported decline in sperm count could be due to observational bias and overinterpretation of linear regression. In fact only 36% of the total variability in sperm count was explained by the linear model and 42% by the quadratic model. The linear model was no longer valid when three new European reports on sperm counts were included in the analysis. The quadratic model, however, suggested an upward trend of sperm count after 1975 (R2 = 0.48, P < 0.0001). Factors other than the 'passage of time' may have contributed to the initial decline of sperm count. An immediate candidate was demography. Our analysis showed that sperm counts in USA were significantly higher in 1938-1956 compared with those in 1957-1974 and 1975-1988, but not in the European or Asian/African/South American countries. The variability of the USA sperm count (1938-1988) explained by the linear and quadratic models was found to be 71 and 70% respectively. The quadratic model importantly indicated that the sperm count in USA decreases asymptotically towards a limiting value and global sperm counts could be increasing since 1970. The non-uniform nature of the global sperm count change suggested that local variations in pollution, diet but not global warming were important determinants of reproductive health.

Demography↗
PubMed1996

Relationship of changes in cardiac output to changes in heart rate in medical ICU patients.

OBJECTIVE: To determine whether changes in cardiac output are correlated with changes in other commonly measured covariables (heart rate, respiratory rate, mean arterial pressure, mean pulmonary artery pressure, pulmonary artery occlusion pressure, and temperature). DESIGN: Case series. SETTING: Medical intensive care unit (ICU) in a Veterans Administration Medical Center. PATIENTS: Twenty-three patients with Swan-Ganz catheters placed by the primary care team were studied on 25 occasions. Patients were managed by the primary team as clinically indicated. INTERVENTIONS: Thermodilution cardiac output and covariables were determined at baseline and at hourly intervals for the next 5 h. Each cardiac output measurement was calculated by averaging the last four of five individual measurements at each time point. RESULTS: The mean cardiac output (9.21/min), heart rate (107/min), and pulmonary artery occlusion pressure (19 mmHg) were elevated. The hourly mean change in cardiac output was 10.2%. Using least-squares linear regression analysis, we found clinically significant changes in cardiac output (> 6.4%) to be most closely correlated with changes in heart rate (R2 = 0.29, p < 0.001). Stepwise linear regression analysis showed that none of the other covariables added significantly to this relationship. No significant relationship was found between changes in cardiac output and changes in pulmonary artery occlusion pressure. Despite these correlations clinically significant changes in cardiac output were accompanied by changes in heart rate in the same direction only 62% of the time. CONCLUSION: Changes in cardiac output were best correlated with changes in heart rate. Changes in pulmonary artery occlusion pressure were not correlated with changes in cardiac output in this population of medical ICU patients. A change in any of the covariables (alone or in combination) cannot be reliably used to indicate a simultaneous change in cardiac output.

Adult↗
PubMed2006

A new formula for correction of total calcium level into ionized serum calcium values in very elderly hospitalized patients.

Ionized calcium (Ca(2+)) seems to be the best measure of active serum calcium but, in France, numerous laboratories do not have Ca(2+) analyzers so that numerous clinicians use Payne's formula to obtain adjusted calcium (Ca(Ad)) values. In frail very elderly patients with protein/energetic malnutrition and very low concentrations of albumin, "correction" with Payne's formula usually gives false hypercalcemic results, so that hypocalcemia may be seriously underdiagnosed. Two hundred and ninety-four patients of 80 years and older with serum albumin level < 35 g/l were included in four French hospitals for elderly people. Biological measurements were standardized in order to determine Ca(2+) and total calcium (Ca(T)) in accordance with approved guidelines. Ca(Ad) was calculated with Payne's formula whereas the dependence of Ca(2+) with serum protein, albumin and Ca(Ad) was investigated by linear regression, the goodness-of-fit of each equation with the measure of Ca(2+) being studied. Taking into account serum protein and albumin levels, multiple linear regression gave the equation: Ca(2+) (mmol/l)=0.188-0.00469 protein (g/l)+0.0110 albumin (g/l)+0.401 Ca(Ad) with r(2)=0.442. The relative difference between the measure and the value given by the equation did not depend upon the center, and the correlation between measured and computed values of Ca(2+) was better, for any group, with our formula than with Payne's formula. When Ca(2+) was expressed with Ca(T) instead of Ca(Ad), albumin term was no longer significant and the new equation was: Ca(2+) (mmol/l)=0.592-0.00449 protein (g/l)+0.410 total calcium (mmol/l) with r(2)=0.438. We propose an alternative to direct measurement of Ca(2+) with a simple formula usable in geriatric units, which are often deprived of high-performance equipment.

Aged, 80 and over↗
PubMed1993

Correlates of cognitive function scores in elderly outpatients.

OBJECTIVE: To determine medical, ophthalmologic, and demographic predictors of cognitive function scores as measured by the Telephone Interview for Cognitive Status (TICS), an adaptation of the Folstein Mini-Mental Status Exam. A secondary objective was to perform an item-by-item analysis of the TICS scores to determine which items correlated most highly with the overall scores. DESIGN: Cross-sectional cohort study. SETTING: The Glaucoma Consultation Service of the Massachusetts Eye and Ear Infirmary. SUBJECTS: 472 of 565 consecutive patients age 65 and older who were seen at the Glaucoma Consultation Service between November 1, 1987 and October 31, 1988. MEASUREMENTS: Each subject had a standard visual examination and review of medical history at entry, followed by a telephone interview that collected information on demographic characteristics, cognitive status, health status, accidents, falls, symptoms of depression, and alcohol intake. RESULTS: A multivariate linear regression model of correlates of TICS score found the strongest correlates to be education, age, occupation, and the presence of depressive symptoms. The only significant ocular condition that correlated with lower TICS score was the presence of surgical aphakia (model R2 = .46). Forty-six percent (216/472) of patients fell below the established definition of normal on the mental status scale. In a logistic regression analysis, the strongest correlates of an abnormal cognitive function score were age, diabetes, educational status, and occupational status. An item analysis using step-wise linear regression showed that 85 percent of the variance in the TICS score was explained by the ability to perform serial sevens and to repeat 10 items immediately after hearing them. Educational status correlated most highly with both of these items (Kendall Tau R = .43 and Kendall Tau R = .30, respectively). CONCLUSION: Education, occupation, depression, and age were the strongest correlates of the score on this new screening test for assessing cognitive status. These factors were stronger correlates of the TICS score than chronic medical conditions, visual loss, or medications. The Telephone Interview for Cognitive Status is a useful instrument, but it may overestimate the prevalence of dementia in studies with a high prevalence of persons with less than a high school education.

Accidental Falls↗
PubMed2003

Fish mercury levels in lakes--adjusting for Hg and fish-size covariation.

Accurate estimates of lake-specific mercury levels are vital in assessing the environmental impact on the mercury content in fish. The intercepts of lake-specific regressions of Hg concentration in fish vs. fish length provide accurate estimates when there is a prominent Hg and fish-size covariation. Commonly used regression methods, such as analysis of covariance (ANCOVA) and various standardization techniques are less suitable, since they do not completely remove the fish-size covariation when regression slopes are not parallel. Partial least squares (PLS) regression analysis reveals that catchment area and water chemistry have the strongest influence on the Hg level in fish in circumneutral lakes. PLS is a multivariate projection method that allows biased linear regression analysis of multicollinear data. The method is applicable to statistical and visual exploration of large data sets, even if there are more variables than observations. Environmental descriptors have no significant impact on the slopes of linear regressions of the Hg concentration in perch (Perca fluviatilis L.) vs. fish length, suggesting that the slopes mainly reflect ontogenetic dietary shifts during the perch life span.

Animals↗
PubMed2005

Factors explaining disability-free life expectancy in Japan: the proportion of older workers, self-reported health status, and the number of public health nurses.

BACKGROUND: Disability-free life expectancy (DFLE) data for 47 prefectures in Japan were reported in 1999; however, few studies have identified the factors associated with the length of the DFLE. The objective of this study was to elucidate the primary factors that explain differences in DFLEs in Japan. METHODS: In our ecological study, 47 prefectures in Japan were used as units of analysis. The DFLEs for men and women at 65 years of age (DFLE65), calculated by Hashimoto et al using Sullivan's method, were set as dependent variables. From various national surveys, 181 factors associated with demographics, socioeconomic status, health status and health behaviors, medical environment, social relationships, climate, and other areas were gathered as independent variables. Pearson's or Spearman's correlation coefficients were calculated to screen independent variables potentially associated with the DFLE65s. Then, multivariate linear regression analyses were conducted for the selected 24 independent variables after adjusting for the proportion of older people (65 years or more) and population density. RESULTS: Multivariate linear regression analyses revealed that the large number of public health nurses per 100,000 population, a good self-reported health status, and a high proportion of older workers were significantly associated with long DFLE65s for both genders. CONCLUSIONS: These three factors could potentially explain the differences in DFLE of the older population in Japan.

Aged↗
PubMed2001

99mTc-DMSA absolute uptake: normal pediatric values at 2-4 hours.

OBJECTIVE: This study aims to determine if normal absolute uptake values of dimercaptosuccinic acid (DMSA) can be predicted accurately over the optimal time for imaging. METHODS: Eighty-eight normal kidneys were analyzed from 44 children with a median age of 4 y. The mean time between injection and scan was 171.7 min. The absolute uptake of DMSA in milligrams was calculated for each kidney. RESULTS: A strong positive linear relationship was found between the absolute uptake of DMSA and the amount of DMSA injected in milligrams (corr = 0.940 [P< 0.0005]), and the age of the child (corr = 0.770 (P<0.0005)). Multiple linear regression showed that these 2 factors accounted for 92.2% of the change in absolute DMSA uptake. CONCLUSION: The absolute uptake of DMSA can be accurately predicted using a linear regression equation incorporating the amount of DMSA injected in milligrams and the patient's age. Time between injection and scan appears to play little role in the absolute uptake of DMSA over 2-4 h.

Adolescent↗
PubMed2001

Ankle muscle stiffness in the control of balance during quiet standing.

This research presents new data and reanalyzed information to refute the criticisms of our model of stiffness control during quiet standing. A re-review of their references to biomechanical research on muscle ankle stiffness confirmed muscle stiffness estimates of the ankle series elastic elements that agreed closely with our estimates. A new technique is presented that directly estimates the muscle stiffness from the ankle moment (N. m) and sway angle (deg). The linear regression of 10 subjects standing quietly for 10 s estimated the stiffness (N x m/deg) to be safely above the gravitational spring. The R(2) scores for this linear regression averaged 0.92, confirming how closely the model approached a perfect spring that would have an R(2) = 1. These results confirm our model of a simple muscle stiffness control and refutes the criticisms.

Adult↗
PubMedDate not supplied

The impact of age-related changes in event-related P300 potentials on detecting early cognitive dysfunction.

Event-related P300 potentials that closely reflect cognitive brain functions show significant age-related latency prolongations. This aging-P300 interaction can best be approximated by third-order polynomial regressions. To delineate the clinical impact this special kind of regression function may have on detecting early cognitive dysfunction, we applied visual P300 potential data of healthy subjects (n = 344; age range, 18-98 years) to nondemented patients with either (i) chronic liver disease (n = 104; age range, 19-74 years) or (ii) cerebral arteriosclerosis (n = 80; age range, 38-80 years). As compared with linear regressions, third-order polynomial regressions for the age-related changes in P300 potential latencies showed a smaller latency increase during middle age, with an accelerating latency prolongation from age 60 onward. In patients with liver cirrhosis, third-order polynomial regressions yielded a rate of abnormal P300 potential latencies exceeding that of linear regressions absolutely by 17-21%, and relatively by 67-71%. Although the rate of P300 abnormalities was much lower in the CAD patients with either regression model, the relative increase in P300 abnormalities due to third-order polynomial regressions was 40-112.5%. In conclusion, normal data for the latencies of P300 potentials based on third-order polynomial regressions result in a higher sensitivity of P300 potentials for detecting early cognitive dysfunction. This gain in diagnostically important information is not offset by a loss in specificity, and may depend on the kind as well as stage of the disease, the age distribution of the patients and the degree of the P300 potential abnormalities.

Journal Article↗
PubMed2005

Regression-based quantitative trait loci mapping: robust, efficient and effective.

Regression has always been an important tool for quantitative geneticists. The use of maximum likelihood (ML) has been advocated for the detection of quantitative trait loci (QTL) through linkage with molecular markers, and this approach can be very effective. However, linear regression models have also been proposed which perform similarly to ML, while retaining the many beneficial features of regression and, hence, can be more tractable and versatile than ML in some circumstances. Here, the use of linear regression to detect QTL in structured outbred populations is reviewed and its perceived shortfalls are revisited. It is argued that the approach is valuable now and will remain so in the future.

Genetics, Population↗
PubMed2004

A quantitative structure-property relationship for predicting drug solubility in PEG 400/water cosolvent systems.

PURPOSE: A quantitative structure-property relationship (QSPR) was developed to predict drug solubility in binary mixtures of polyethylene glycol (PEG) 400 and water. The ability of the QSPR model to predict solubility was assessed and compared to the classic log-linear cosolvency model. METHODS: The solubility of 122 drugs, ranging in log P from -2.4 to 7.5, was determined in 0%, 25%, 50%, and 75% PEG (v/v in water) by the shake-flask method. Solubility data from 84 drugs were fit by linear regression using the following molecular descriptors: molecular weight, volume, radius of gyration, density, number of rotatable bonds, hydrogen-bond donors, and hydrogen-bond acceptors. The multiple linear regression model was optimized by a genetic algorithm guided selection method. The remaining 38 compounds were used to test the predictability of the model. RESULTS: QSPR-based models developed at each volume fraction with the training set compounds showed a reasonable correlation coefficient (r) of approximately 0.9 and a root mean square (rms) error of <0.5 log unit. The model predicted solubility values of approximately 78% of the testing set compounds within 1 log unit. The log-linear model was as effective as the QSPR-based model in predicting the testing set solubilities; however, many drugs, as expected, showed significant deviation from log-linearity. CONCLUSIONS: The QSPR model requires only the chemical structure of the drug and has utility for guiding vehicle identification for early preclinical in vivo studies, especially when compound availability is limited and experimental data such as aqueous solubility and melting point are unknown. When experimental data are available, the log-linear model was verified to be a useful predictive tool.

Models, Chemical↗
PubMed2001

Measurement of single kidney contrast media clearance by multiphasic spiral computed tomography: preliminary results.

OBJECTIVE: We present preliminary results of a new method (hereinafter called 'CT-clearance') to measure single kidney contrast media clearance by performing multiphasic helical CT of the kidneys. CT-clearance was calculated according to an extension of the Patlak-Plot. In contrast to prior investigators, who repeatedly measured a single slice, this method makes it possible to calculate single kidney clearance from at least three spiral CTs, utilizing the whole kidney volume. METHODS: Spiral CT of the kidneys was performed unenhanced and about 30 and 100 s after administration of about 120 ml iopromide. Sum-density of the whole kidneys and aortic density was calculated from this data. Using this data, renal clearance of contrast media was calculated by CT-clearance in 29 patients. As reference, Serum-clearance was calculated in 24 patients by application of a modified one-exponential slope model. Information on the relative kidney function was gained by renal scintigraphy with Tc99m-MAG-3 or Tc99m-DMSA in 29 patients. RESULTS: Linear regression analysis revealed a correlation coefficient of CT-clearance with Serum-clearance of r=0.78 with Cl (CT) [ml/min]=22.2+1.03 * Cl (serum), n=24. Linear regression of the relative kidney function (rkf) of the right kidney calculated by CT-clearance compared to scintigraphy results provided a correlation coefficient r=0.89 with rkf(CT)[%]=18.6+0.58 * rkf(scintigraphy), n=29. CONCLUSION: The obtained results of contrast media clearance measured by CT-clearance are in the physiological range of the parameter. Future studies should be performed to improve the methodology with the aim of higher accuracy. More specifically, better determination of the aortic density curve might improve the accuracy.

Contrast Media↗
PubMed1980

[Quantimetric investigations of the time of death by estimating the postmortem threshold (rheobase) of human skeletal muscles to electric stimulus by direct current (author's transl)].

We recently reported on an unusually high correlation between the post mortem interval and the postmortem (pm) increase of the muscular threshold to galvanic stimuli (Rheobase) in the skeletal muscle of the rat. To find out whether a practicable method of estimating the time of death could be developed out of this, we investigated skeletal muscles of 11 individuals having died from various causes. The muscles were stimulated electrically by square-impulses of 1 s length at 200 V, the strength ranging from 0.1 to 80 mA. The electrodes were applied to the skin. A strong correlatin (r = 0.98 on average) between the muscular threshold to electric stimuli and the time of pm investigation was also found in human muscles and can be described by the Formula Y(T)[mA] = A*Exp(b . T) when A* is the primary threshold (Rheobase) and T the time of pm interval. Differentiation of this equation leads to a strongly linear regression curve by which finally the time of death can be expressed as (formula: see text). The coefficients a0 and b can be calculated by the linear regression of the pm Rheobase in a semi-logarithmic system of coordinates and a* as lnA* [mA] from the course of the regression curves--a0 = lnA0 [mA] at the beginning of the investigation and b [l/min] = gradient of the regression. The testing of the muscles flexor digiti V and flexor digitorum communis brevis, in particular, lead to the most exact results. By this method immediate results of hitherto unknown reliability can be achieved in estimating the time of death up to 15 h postmortem by statistical means of calculation. Changing temperatures, different causes of death, and other individual factors seem to be of no influence on the results so that collectives of known characteristics are not necessary for comparison.

Adult↗
PubMed2005

QSPR modeling of pseudoternary microemulsions formulated employing lecithin surfactants: application of data mining, molecular and statistical modeling.

Data mining, computer aided molecular modeling, descriptor calculation, genetic algorithm and multiple linear regression analysis techniques were combined together to generate predictive quantitative structure property relationship (QSPR) models explaining the formation of lecithin-based W/O microemulsions. Ninety-four microemulsion phase diagrams were collected from five different references published over the past few years. Computer-based molecular modeling techniques were then applied on the components of the collected microemulsion systems to generate corresponding plausible three-dimensional (3D) structures. The resulting 3D models were utilized to calculate a group of molecular physicochemical descriptors. Thereafter, genetic algorithm and backward stepwise regression analysis were separately assessed as means for selecting optimal descriptor sets for statistical modeling. The selected descriptors were correlated with microemulsion existence areas employing multiple linear regression analysis. The resulting W/O models were statistically validated and found to be of significant predictive power. The models allowed better understanding of the process of microemulsion formation. Unfortunately, all QSPR modeling efforts directed towards O/W microemulsions failed completely.

Emulsions↗
PubMed1990

Progress-curve equations for reversible enzyme-catalysed reactions inhibited by tight-binding inhibitors.

The rate equation for a tight-binding inhibitor of an enzyme-catalysed first-order reversible reaction was used to derive two integrated equations. One of them covers the situations in which competitive, uncompetitive or non-competitive inhibition occurs and the other refers to the special non-competitive case where the two inhibition constants are equal. For these equations, graphical and non-linear regression methods are proposed for distinguishing between types of inhibition and for calculating inhibition constants from progress-curve data. The application of the non-linear regression to the analysis of stimulated progress curves in the presence of a tight-binding inhibitor is also presented. The results obtained are valid for any type of 'dead-end'-complex-forming inhibitor and can be used to characterize an unknown inhibitor on the basis of progress curves.

Binding, Competitive↗