SEARCH · Search PubMed

Results for “Statistical 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.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 127 records · Page 7Linked to original sources

PubMed2003

QSAR modeling of anti-HIV activities of alkenyldiarylmethanes using topological and physicochemical descriptors.

Three series of anti-HIV data (reverse transcriptase inhibitory activity, cytopathicity data, and cytotoxicity data) of alkenyldiarylmethanes were modeled with physicochemical, topological and structural descriptors by multiple regression analysis using principal component factor analysis as the data pre-processing step. Molar refractivity was found to be a significant contributor in modeling all three data sets. Apart from this, partition coefficient, E-state index, valence connectivity and indicator parameters were important in modeling different activity series. The final relations were of moderate to good quality as evidenced from regression statistics (R2 values ranging 66-75%) and leave-one-out cross validation data (Q2 values ranging 54-70%).

Alkenes↗

PubMed2004

A longitudinal study of dehydroepiandrosterone sulphate (DHEAS) change in older men and women: the Rancho Bernardo Study.

OBJECTIVE: To examine the age-related and sex-specific rates and determinants of change in endogenous dehydroepiandrosterone sulphate (DHEAS) levels in older community-dwelling adults, taking into account regression to the mean. DESIGN: Prospective cohort study from Rancho Bernardo, California, USA. METHODS: Plasma for DHEAS was collected at baseline and 10-14 years later from 242 men and 207 postmenopausal women (age range, 60-77 years), who were not taking hormone therapy or corticosteroids. Age-related and sex-specific changes in DHEAS were calculated according to four different definitions of change: absolute change, annual percentage change, change exceeding 10% of baseline and change exceeding that expected from statistical regression to the mean. Determinants of DHEAS change were assessed using regression analyses. RESULTS: Baseline DHEAS levels were higher for men than women (age-adjusted means, 1.37+/-0.73 microg/ml (3.72+/-1.98 micromol/l) vs 0.73+/-0.48 microg/ml (1.98+/-1.30 micromol/l) respectively, P<0.0001), with more pronounced declines observed in men (-4.0%/year) compared with women (-2.1%/year; P<0.001). Some 28% of women and 5% of men had DHEAS levels that increased or stayed the same according to a 10% definition of change. When regression to the mean artifact was accounted for, only 15% of women and 5% of men showed true increases in DHEAS. In both sexes, baseline DHEAS levels accounted for three-quarters of the variability in absolute DHEAS change over time, with higher baseline levels resulting in greater loss. Sex, baseline weight, age and smoking status were significantly associated with DHEAS change in univariate models; only sex remained independently associated with DHEAS change in multivariate analyses, with men showing greater annual declines than women (estimated coefficient=0.006, P=0.008). CONCLUSION: In this sample, over 30% of the observed changes in DHEAS could be attributed to regression to the mean. Potential underlying mechanisms of change in DHEAS levels, and sex differences, are discussed.

Aged↗

PubMed1998

An analysis of mortality in the grower/finisher phase of swine production in the United States.

Over a 6-month period, the mean mortality risk (based on 393 operations participating in the United States National Animal Health Monitoring System 1995 National Swine Study, and representing operations with > or = 300 market hogs in 16 states), was 2.3 +/- 0.2% in the grower/finisher production phase (where figures after the +/- represent the standard error of the estimate). Mortality > or = 4% was experienced by 13.5 +/- 2.9% of grower/finisher operations, while 63.6 +/- 5.3% had < or = 2% mortality. To identify factors associated with > or = 4% mortality, stepwise logistic regression [Statistical Analysis Systems, 1989. SAS/STAT User's Guide, Version 6, 4th edn, Vol. 2. SAS Institute, Cary, NC, 794 pp.] was performed twice: once using operations with all mortality rates, and again excluding operations with between 2% and 4% mortality. Final models were run with SUDAAN [Shah, B.V., Barnwell, B.G., Bieler, G.S., 1996. SUDAAN User's Manual, Version 6.40, 2nd edn. Research Triangle Institute, Research Triangle Park, NC, 492 pp.] to take the sample design into account. In addition, SAS and SUDAAN logistic regression models were developed to analyze factors associated with > 2.3% mortality among grower/finisher pigs. Mean weaning age < or = 28 days entered all models as being associated with increased mortality in the grower/finisher unit. Not obtaining all grower/finisher pigs from farrowing units belonging to the operation was associated with > or = 4% mortality among grower/finisher swine. Not typically giving grower/finisher pigs antibiotics or other agents as disease-preventives or growth-promotants in the feed or water, and ranking producer organizations as very or extremely important sources of antibiotic information were associated with < or = 2.3% mortality in the grower/finisher phase.

Aging↗

PubMed2001

Introduction to artificial neural networks for physicians: taking the lid off the black box.

BACKGROUND: Over the past 5 years, a steady stream of publications has discussed the use of artificial neural networks (ANNs) for urologic and other medical applications. The pace of this research has increased recently, and deployed products based on this technology are now appearing. Before these tools can be widely accepted by clinicians and researchers, a deeper level of understanding of ANNs is necessary. This article attempts to lay some of the groundwork needed to facilitate this familiarity. METHODS: A short discussion of neural network history is included for background. This is followed by an in-depth discussion of how and why ANNs work. This discussion includes the relationship between ANNs and statistical regression. An investigation of issues associated with neural networks follows, applicable to both general and urologic-specific applications. RESULTS: Neural networks are computer models that have been studied extensively for over 50 years, with prostate cancer applications since 1994. From a biological viewpoint, ANNs are artificial analogues of data structures that exist in nervous systems. From a numeric viewpoint, ANNs are matrices of numbers whose values comprise knowledge that is distilled from historic databases. Many types of neural networks are analogous to well-known statistical methods. CONCLUSIONS: ANNs are complex numeric constructs, but no more complex than similar statistical methods. However, several issues associated with neural network derivation demand that developers apply rigorous engineering practices in their studies.

Humans↗

PubMed1983

Stratification of risk factors in renal cell carcinoma.

One hundred fifteen patients underwent surgical treatment of renal cell carcinoma. Survival time was analyzed by statistical regression methods in order to determine the joint significance of stage, using both the Robson and TNM classifications, histologic grade, cell type, and demographic features of the patient sample. Grade and cell type were essentially interchangeable with respect to predicting survival in patients with and without metastatic disease. The strong association between these two factors explains and supports this result. The presence of metastases dramatically altered survival. Local extent of the tumor was an important indicator of survival in the nonmetastatic group of patients, but was not a statistically significant factor in those patients presenting with widespread disease. Venous involvement was not found to be of prognostic significance in either group. Age and weight loss contributed to predicting survival in the nonmetastatic patient group in addition to the measures of disease extent already discussed. Sex and race were not significant indicators of length of survival. In the group of patients with metastatic disease, no difference in survival was observed between soft tissue and bony metastases.

Adenocarcinoma↗

PubMed1986

Methodology for statistical analysis of SENCAR mouse skin assay data.

Various response measures and statistical methods appropriate for the analysis of data collected in the SENCAR mouse skin assay are examined. The characteristics of the tumor response data do not readily lend themselves to the classical methods for hypothesis testing. The advantages and limitations of conventional methods of analysis and methods recommended in the literature are discussed. Several alternative response measures that were developed specifically to answer the problems inherent in the data collected in the SENCAR bioassay system are described. These measures take into account animal survival, tumor multiplicity, and tumor regression. Statistical methods for the analysis of these measures to test for a positive dose response and a dose-response relationship are discussed. Sample data from representative initiation/promotion studies are used to compare the response measures and methods of analysis.

Animals↗

PubMed2006

Coexistence of angiotensin II type-1 receptor A1166C and angiotensin-converting enzyme D/D polymorphism suggests susceptibility for small-vessel-associated ischemic stroke.

The renin-angiotensin system plays an important role in the maintenance of blood pressure homeostasis. The angiotensin-converting enzyme (ACE) converts angiotensin I into angiotensin II. Angiotensin II, which binds the angiotensin II type-1 receptor (AT1R), is a potent vasoconstrictor. On a pathophysiological basis, both ACE I/D and AT1R A1166C polymorphism lead to an enhanced activity of the angiotensin II-AT1R axis, thereby possibly contributing to circulatory disturbances. A mutually facilitatory effect may be presumed between the two polymorphisms. We examined whether this synergistic effect is involved in the evolution of different types of ischemic stroke. Genetic and clinical data on 308 consecutive patients with acutely developing ischemic stroke were analyzed. Atotal of 272 stroke and neuroimaging alteration-free subjects served as a control group. Univariate and logistic regression statistical approaches were used. The ACE D allele combined with the AT1R 1166C allele did not yield a risk of ischemic stroke. However, the co-occurrence of the homozygous ACE D/D and at least one AT1R 1166C allele was more frequent in the ischemic stroke group than in the control group (22.4 vs 11%, p < 0.005, OR, 2.33; 95% CI, 1.46-3.7). After specific subgroup analysis, this synergistic association was even stronger for small-vessel ischemic stroke (OR, 3.44; 95% CI, 1.9-6.24; p < 0.0005). Multivariate logistic regression analysis of the data confirmed this association (adjusted OR, 3.54, 95% CI, 1.88-7.16; p < 0.0005). Our results demonstrate that ACE D/D and AT1R 1166C polymorphism were associated with the development of small-vessel ischemic stroke through a mutually facilitatory interplay between them. Genetic interactions might contribute to the altered functional network in renin-angiotensin system in vascular disorders.

Adult↗

PubMedDate not supplied

Can physique and gluteal size predict penile length in adult Nigerian men?

BACKGROUND: In Nigeria, especially among the Igbo tribe, there is a well-held belief that a man's penile size can be predicted from his physique and the size of his buttocks, with people of small physique and flat buttocks likely to have long penile lengths. STUDY DESIGN: A prospective study to test the scientific veracity of this traditional and apparently mythical belief. Stretched penile length was measured in 115 men between the ages of 30-65 years and its correlation with the body-mass index and the circumference of the hip as measured around the most prominent points on their buttocks, was statistically determined. RESULT: The mean age of the subjects was 42.30 years (SD = 9.67), with a median of 40 years and a range of 30-65 years. The mean penile length was 13.37 cm with a median of 13 cm and a range of 7.5-19.5 cm. The mean circumference of the body around the buttocks was 96.46 cm (SD = 10.91), median 98 cm, and range 73-122 cm. The body-mass index ranged 17.34-44.44, with the mean at 26.87 (SD = 5.86), and the median 25.53. Linear regression statistics showed no statistically significant correlation between stretched penile length and body-mass index, thus physique. There was a significant direct correlation between penile length and gluteal size. CONCLUSION: The supposed relationship between penile length and gluteal size may have a scientific basis, but contrary to belief, large buttocks is more predictive of longer penile length than small buttocks. Penile length has no relationship to physique.

Adult↗

PubMed1994

Regression of lateral asymmetry scores toward the mean.

On most, if not all, laterality tasks, some subjects initially display a left sensory field advantage and others a right sensory field advantage. During the course of testing, the mean asymmetry score of those subjects who start with a left sensory field advantage decreases, whereas the mean asymmetry score of those subjects who start with a right sensory field advantage increases (asymmetry scores computed as L-R). Some previous studies (e.g., Ross-Kossak and Turkewitz, 1986) have suggested that these differential, temporal shifts in sensory field advantage reflect systematic changes in subjects' sensory field advantage (and associated hemispheric advantage) during familiarization and individual differences in those changes. Alternatively, the differential, temporal shifts may reflect unsystematic sources of changes because they are not distinguishable from statistical regression toward the mean. Consistent with this hypothesis, "time-reversed control" analyses of laterality data show that the differential trends do not reflect a factor associated with the specific direction of time passage as would be predicted in the case of increasing stimulus familiarization. Regression of asymmetry scores toward the mean, or its equivalent, imperfect reliability of asymmetry scores, may be significantly attributable to short-term fluctuations in subjects' hemispheric arousal asymmetries.

Adult↗

PubMed2000

Cardiopulmonary resuscitation skills in nurses and nursing students.

The purpose of this study is to describe the basic cardiac life-support (BLS) skills of nurses and nursing students in southern Finland and Hungary, and to assess the influence of resuscitation teaching and other group characteristics on performance. The data for the study were collected in the spring and autumn of 1997. The study group consisted of 75 nurses from Helsinki University Central Hospital's medical outcome unit, 188 final term students in four nursing institutes in Uusimaa county and 35 final term students in a Hungarian institute of nursing. A total of 298 people (34 men and 264 women) participated in the study. Background information was collected using a structured questionnaire devised specifically for this study. Resuscitation skills were measured using the Skillmeter Anne manikin. The manikin was placed supine during the test. After completing the questionnaire, every participant attempted resuscitation on the manikin - which was supplied with a printer - for 4 min. The results were printed out and attached to the questionnaires. The data were analysed using two-way frequency tables and logistic regression. Statistical differences were calculated using the chi(2)-test. The results showed that 53% of the participants had studied resuscitation during the last 6 months, but 7% had never participated in resuscitation teaching. Before testing, 55% of the participants estimated that their resuscitation skills were good. The results showed that 36% first assessed the patient's response, 67% opened the airway but only 3% determined pulselessness before starting to resuscitate. Twenty-one percent of the participants compressed correctly for at least half of the test and 33% ventilated correctly at least half of the time. Logistic regression showed that the best predictors for good response assessment skills went to those who were nursing students who had studied resuscitation skills sometime during the previous 6 months. The best predictor of the skill to open the airway was a positive attitude towards personal cardiopulmonary resuscitation (CPR) skills, i.e. self-confidence. The predictor for adequate skills in artificial ventilation was that they belonged in the group of nursing students who had benefited from recent resuscitation training (<6 months). In conclusion, the skills of the participants of the study can not be considered adequate in terms of an adequate and prompt assessment of the need for resuscitation, and a 50% success rate in artificial ventilation and chest compression.

Adult↗

PubMed2004

Prediction of 6-month survival after liver transplantation using Cox regression.

BACKGROUND: The outcome of liver transplantation (LTx) has been correlated with several donor and recipient factors. METHODS: A database of 191 consecutive LTx cases was analyzed using Kaplan-Meier and Cox regression statistics based on 80 variables. To avoid additional effects of late events on patient survival, the chosen endpoint was 6 months. Data were evaluated using SPSS statistical software. RESULTS: Kaplan-Meier analysis revealed a difference in 1- to 6-month graft survival between patients transplanted with organs from donors older versus younger than 60 years (Breslow, P <.01). Differences in 1- to 6-month graft survivals were observed between patients listed as UNOS status 3, 2B, 2A, and 1: the outcomes for UNOS status 2B versus UNOS status 2A and UNOS status 2B versus status 1 were significant (P <.05). Differences in 1- to 6-month graft survival rates were found between patients with versus without sepsis (P <.05), and with versus without rejection episodes (P <.01). Cox regression analysis revealed only three of the variables to be independent prognostic predictors of graft failure: donor age; postoperative septic status; and rejection. The best mathematical multivariate Cox regression model linked donor age + donor Na + rejection + sepsis to 1- to 6-month graft survival (chi-square = 29.06, P <.001). CONCLUSION: Factors predictive of 1- to 6-month graft survival after liver transplantation include donor age; UNOS status; sepsis; and rejection.

Age Factors↗

PubMed1997

Costs of poison-related hospitalizations at an urban teaching hospital for children.

BACKGROUND: Childhood poisonings continue to exact a notable toll in injury-related morbidity and economic cost. Because a substantial portion of this morbidity and economic cost is associated with hospitalization, an analysis of the cost of poisoning hospitalizations might help to identify areas in which medical care could be improved and costs could be reduced. OBJECTIVE: To assess the aggregate cost of poison-related hospitalizations and to analyze the trends in categorical poisoning costs during a 4-year period. DESIGN: Cost-benefit analysis of charge data and length of stay (LOS) for poison-related hospitalizations. SETTING: Admissions to one urban children's hospital. PATIENTS: Case mix data were reviewed to identify those children whose hospitalizations had a primary discharge diagnosis related to poisoning in fiscal years 1992 to 1995. MAIN OUTCOME MEASURES: Mean aggregate hospital charges and LOS for poisonings were compared for each of the 4 years of the study. Comparisons were also made with charges and LOS for hospitalizations for all other diagnoses during the same period. Trends in hospital charges and LOS for the most common specific types of poisoning were analyzed individually. Linear regression statistics were used to compare the costs and LOS of specific types of poisoning. RESULTS: There were 638 poison-related hospitalizations during the 4-year study (0.9% of all pediatric hospital admissions). Charges per case decreased from $7934 in fiscal year 1992 to $4968 in fiscal year 1995 (z = -2.74, P = .006); mean LOS decreased from 5.85 days in 1992 to 3.45 days in 1995 (z = 2.84, P = .005). These trends exceeded smaller trends toward decreasing charges and LOS seen also for non-poison-related hospitalizations. Acetaminophen, lead, and antidepressant medications were the most common and most costly specific agents implicated in poison-related hospitalizations during the study period. Linear regression analysis showed a significant decrease in charges (F = 6.35, R2 = 0.09, P = .014) and LOS for acetaminophen (F = 4.30, R2 = 0.063, P < .04) but not for lead or antidepressant poison-related hospitalizations. CONCLUSIONS: Despite an increasing number of children hospitalized for poisoning at one institution during a 4-year period, per case hospital charges decreased substantially. The cost savings were associated with a marked decrease in LOS. Still, poison-related admissions to a single pediatric facility accounted for almost $1 million in hospital charges in fiscal year 1995. A few agents are overrepresented; new poisoning prevention measures aimed at these toxic agents are warranted. We conclude that further outcomes studies are needed to delineate cost-effective improvements in patient care targeted toward poisonings owing to those agents.

Adolescent↗

PubMed1992

Phase shift in leaf movements of xanthium attributed to age and rhythm patterns.

Leaves of cockelbur (Xanthium strumarium L.) have been reported to be in either an upright or downward position during the dark span (night) of a 24-hour cycle. Results from our studies clearly indicate that such differences in leaf position are not related to differences in ecotypes but can be attributed to age of the leaf, pattern of the waveform of the rhythm at various stages of the light-dark synchronizer regimen, and the statistical model used for the analysis of the waveform. Younger leaves reached a maximum upright position closer to the middle of the dark span, whereas older leaves reached this position closer to the end of the dark span. A phase shift of up to 6 to 10 hours may occur as the leaf ages. Results from the examination of the pattern of the waveform at four different times showed that the pattern of a younger leaf was different from that of an older leaf during the middle of the dark span, during the light-to-dark transition, and during the middle of the light span, but not during the dark-to-light transition. Linear regression, statistical analyses, and the fitting of harmonics clearly indicate that it is the trough, more than the peak, that differs with the age of the leaf.

Journal Article↗

PubMed1988

Risk factors for cancer of the renal parenchyma.

A case-control study of cancer of the renal parenchyma (360 cases and 985 controls) investigated lifestyle and dietary risk factors by means of a questionnaire survey. A lifetime history concerning urological disease and consumption of analgesics, tobacco, and prescribed medicines, was obtained together with current dietary and demographic information. Estimates of relative risk were calculated using logistic regression. Statistically significant associations were found with phenacetin-containing analgesics, urological disease and increasing consumption of milk. No association was found with consumption of aspirin, paracetamol, prescribed medicines, tea, coffee or animal proteins other than milk, or with the type of fat or oil used for cooking or spreading. The association with tobacco just failed to reach significance.

Australia↗

PubMed1980

Estrogen, androgen, glucocorticoid, and progesterone receptors in progestin-induced regression of human breast cancer.

A study was made of basic mechanisms involved in regression of breast cancer exposed to high levels of synthetic progestins. The possibility that progestins act on breast cancer by way of the progesterone receptor mechanism and subsequent increase of estradiol 17 beta-dehydrogenase activity could not be confirmed in this investigation. It is demonstrated that the progestins megestrol acetate and medroxyprogesterone acetate are strong competitors for steroids which bind specifically to androgen, glucocorticoid, and progesterone receptors, indicating that the progestins are able to bind to these receptors with high affinity. In contrast, these progestins do not compete with estradiol for estrogen receptor binding. In 34 patients with progressive metastatic breast cancer, results of receptor studies have been correlated with clinical response during treatment with megestrol acetate. Statistically, regressions were significantly associated with tumors containing large amounts of androgen receptors. Clinical correlation with the quantities of glucocorticoid receptor was weak, while such correlations with estrogen and progesterone receptors were absent. However, we did demonstrate relationships between the quantities of the various receptors in breast cancer. Tumors containing a large amount of androgen receptors also generally contain estrogen receptors. It might be that a favorable response to progestins is confined to the group of patients with hormone-responsive breast cancers, as such characterized by the presence of estrogen receptors, and that within this group the actual androgen receptor levels determine response.

Breast Neoplasms↗

PubMed1995

Further comments on the estimation of error associated with the Gustafson dental age estimation method.

Many researchers in the field of forensic odontology have questioned the error estimates stated in Gustafson's paper outlining the relationship between certain dental attributes and age. In a substantial re-working of Gustafson's data, Maples and Rice corrected Gustafson's regression statistics and found that the error associated with the age estimate was nearly double that claimed by Gustafson. We offer another statistical analysis of Gustafson's data and find that the errors calculated by Maples and Rice were also in error, being about a year too small. We give a formula for what we believe to be the correct treatment of errors in such cases, but conclude by observing that there is an urgent need for a more rigorous study of the traits first tabulated by Gustafson.

Adolescent↗

PubMed1979

Statistical models of the demand for emergency medical services in an urban area.

First- and second-order statistical regression models are presented for the Emergency Medical Services (EMS) demand in an urban area as it relates to various socioeconomic, demographic, and other characteristics of the area. Individual models are formulated for different types of medical emergencies with the city of Atlanta, GA, serving as the data base. These models are generally shown to provide excellent fits to the empirical data.

Emergency Medical Services↗

PubMed2006

Partial least squares: a versatile tool for the analysis of high-dimensional genomic data.

Partial least squares (PLS) is an efficient statistical regression technique that is highly suited for the analysis of genomic and proteomic data. In this article, we review both the theory underlying PLS as well as a host of bioinformatics applications of PLS. In particular, we provide a systematic comparison of the PLS approaches currently employed, and discuss analysis problems as diverse as, e.g. tumor classification from transcriptome data, identification of relevant genes, survival analysis and modeling of gene networks and transcription factor activities.

Animals↗