Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Logistic Model”

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 1,423 records · Page 79Linked to original sources
PubMed2006

Risk factors for Clostridium difficile toxin-positive nosocomial diarrhoea.

Data were retrieved from the records of all patients from whom stool was sent for Clostridium difficile toxin testing during the year 2001. Toxin-positive and -negative patients were compared by bivariate analysis and regression models. Eight hundred samples from 610 patients were sent for C. difficile toxin testing. Charts of 535 patients (88%) were available for analysis. Of those, 17% had a positive toxin test whilst 83% had a negative toxin test. There was no difference in the number of daily bowel movements between the two groups. Toxin-positive patients were older (P<0.0001), more often came from nursing homes (P<0.05), had higher leukocyte counts (P<0.001), higher blood urea nitrogen (P<0.01), lower serum albumin (P<0.01) and more often received diuretics (P<0.01) and clindamycin (P<0.05). Logistic regression analysis showed that previous antibiotic-associated diarrhoea was the most significant risk factor for toxin-positive diarrhoea (P<0.001), followed by clindamycin treatment (P<0.005), diuretics (P<0.005) and older age (P<0.05). Another logistic model showed the contribution of macrolides (P<0.05) to the development of hospital-acquired diarrhoea.

Adolescent↗
PubMed2002

Neurological outcomes in coronary surgery: independent effect of avoiding cardiopulmonary bypass.

BACKGROUND: Recent studies examining neuroprotective effects of off-pump coronary artery bypass grafting (CABG) have shown inconsistent results. We examined our database to quantify the independent effects of avoidance of cardiopulmonary bypass (CPB) and aortic manipulation on neurologic outcomes after CABG. METHODS: A total of 2,327 consecutive cases undergoing isolated CABG between April 1997 and May 2001 were identified at our two institutions. Patients were divided into three groups: on CPB, off-pump with aortic manipulation, and off-pump without aortic manipulation. To control for the confounding effects of other risk factors, we performed a multivariate logistic regression analysis. Potential covariates considered in the logistic model included age, sex, redo operations, diabetes, chronic obstructive pulmonary disease, neurologic disease, peripheral vascular disease, ejection fraction, and priority of operation. RESULTS: A total of 1,210 cases were performed on CPB, compared with 520 off-pump with aortic manipulation, and 597 off-pump without aortic manipulation. The incidence of focal neurologic deficit was 1.6% (n = 19) in the on-pump group, 0.4% (n = 2) in the off-pump with aortic manipulation group, and 0.5% (n = 3) for the off-pump without aortic manipulation group (p for trend = 0.027). The results of the multivariate logistic regression analysis demonstrated that use of CPB was a risk factor for focal neurologic deficit, with an odds ratio of 3.82 (95% confidence interval, 1.41 to 10.34; p = 0.005). Aortic manipulation did not significantly influence neurologic outcome in off-pump patients. CONCLUSIONS: Off-pump operation, with or without aortic manipulation, reduces adverse neurologic outcomes compared with on-pump procedures.

Aged↗
PubMed2006

Myeloperoxidase aids prognostication together with N-terminal pro-B-type natriuretic peptide in high-risk patients with acute ST elevation myocardial infarction.

BACKGROUND: Inflammation plays a critical role in acute myocardial infarction (MI). One such inflammatory marker is myeloperoxidase (MPO). Its role as a predictor of death or MI in patients with ST segment elevation myocardial infarction (STEMI) is unclear. AIM: To investigate the role of MPO as a predictor of death or MI in patients with STEMI and to compare it with N-terminal pro-B-type natriuretic peptide (NT-BNP). METHOD: 384 post STEMI patients were studied. Patients were followed up for the combined end point of death or readmission with non-fatal MI. RESULTS: There were 40 deaths and 37 readmissions with MI. Median MPO was raised in patients experiencing death or MI than in survivors (median (range), 50.6 (15.3-124.1) ng/ml vs 33.5 (6.6-400.2) ng/ml, p = 0.001). Using a Cox proportional hazards model, log median MPO (HR 6.91, 95% CI 1.79 to 26.73, p = 0.005) and log median NT-BNP (HR 4.21, 95% CI 1.53 to 11.58, p = 0.005) independently predicted death or non-fatal MI. MPO had predictive power in both below and above median NT-BNP levels (log rank 5.60, p = 0.020 and log rank 5.12, p = 0.024, respectively). The receiver-operating curve for median NT-BNP yielded an area under the curve (AUC) of 0.72 (95% CI 0.65 to 0.79, p<0.001); for median MPO, the AUC was 0.62 (95% CI 0.55 to 0.69, p = 0.001). The logistic model combining the two markers yielded an AUC of 0.76 (95% CI 0.69 to 0.82, p<0.001). CONCLUSION: MPO and NT-BNP may be useful tools for risk stratification of all acute coronary syndromes, including patients with STEMI at higher risk.

Aged↗
PubMed1997

Associations between carotid atherosclerosis and high factor VIII activity, dyslipidemia, and hypertension.

BACKGROUND AND PURPOSE: A subsample of 147 Chinese subjects from a population-based study of cardiovascular diseases (Cardiovascular Disease Risk Factor Two-Township Study) participated in an ancillary study on extracranial carotid color duplex ultrasonography that aimed to assess the relations of coagulation factors to stroke and carotid atherosclerosis. METHODS: Logistic models were used to study the associations between cardiovascular disease risk factors and stroke/carotid atherosclerosis, controlling for the effects of age and sex. RESULTS: Stroke was significantly associated with hypertension and high values of plasma glucose but not with fibrinogen, factor VIIc, or factor VIIIc. Carotid plaques identified in this study were mostly mild and moderate. The presence of these mild and moderate carotid plaques was significantly associated with high values of factor VIII activity, hypercholesterolemia, hypertriglyceridemia, and hypertension. The highest tertile of factor VIIIc (> 1.53) was associated with an odds ratio of 3.35 for carotid atherosclerosis when compared with the lowest tertile (< 1.20). A multiple logistic regression including all significant risk factors showed that the degree of association between factor VIIIc and atherosclerosis was attenuated to an odds ratio of 2.65 (P = .061). CONCLUSIONS: In the present study, the roles of hypertension, hypercholesterolemia, and hypertriglyceridemia have been implicated in the pathogenesis of carotid atherosclerosis, and roles for hypertension and hyperglycemia in stroke were indicated. A positive association between factor VIIIc and carotid atherosclerosis in this Chinese population was found. Whether this association is independent of the effect of other cardiovascular risk factors awaits further study.

Adult↗
PubMed1985

Testing for a trend in tumor prevalence rates: I. Nonlethal tumors.

In the analysis of animal carcinogenicity studies, the standard survival-adjusted test for a dose-related trend in the prevalence of nonlethal tumors is the Hoel-Walburg test, which stratifies on age at death by grouping survival times into intervals. An alternative analysis assesses trend on the basis of the likelihood score test under a logistic model for the prevalence function, which adjusts for survival by including age at death as a continuous regression variable. Extensive simulations demonstrate that the test based on modeling the prevalence log-odds as a linear function of age is more powerful than the Hoel-Walburg test, regardless of the intervals used by the latter to stratify the data. Without incorporating a continuity correction, the size of each test often exceeds the nominal level, especially when the mortality patterns differ across dose groups. Corrected versions of the tests operate at conservative levels, where the degree of conservatism varies with the distribution of the data. When the mortality patterns for the dose groups are similar, both tests have essentially the same power to detect a trend in tumor prevalence rates. However, when mortality varies with dose, the logistic regression test with a linear age term is more powerful than the Hoel-Walburg test, and this gain in power increases as the dose-specific mortality patterns become more disparate.

Aging↗
PubMed1977

Time delays in single species growth models.

A general model is considered for the growth of a single species population which describes the per unit growth rate as a general functional of past population sizes. Solutions near equilibrium are studied as function of epsilon = 1/b, the reciprocal of the inherent per unit growth rate b of the population in the absense of any density constraints. Roughly speaking, it is shown that for large epsilon the equilibrium is asymptotically stable and that for epsilon small the solutions show divergent oscillations around the equilibrium. In the latter case a first order approximation is obtained by means of singular perturbation methods. The results are illustrated by means of a numerically integrated delay-logistic model.

Models, Theoretical↗
PubMed2003

Predictive factors of 5-year health assessment questionnaire disability in early rheumatoid arthritis.

OBJECTIVE: To determine prognostic factors of disability in early rheumatoid arthritis (RA) and to investigate the radiological and functional course of the disease. METHODS: A total of 191 patients with early RA (diagnosed for less than one year) according to American College of Rheumatology criteria were followed prospectively for 5 years. At baseline and at endpoint, Stanford Health Assessment Questionnaire (HAQ) scores and radiological scores (Sharp's score modified by van der Heijde) were performed. Correlations between numerous baseline data and HAQ score at endpoint were analyzed, using nonparametric tests. A multilinear regression model was performed to select independent prognostic factors of HAQ disability. RESULTS: During the 5-year followup, mean HAQ decreased from 1.3 (+/- 0.7) to 0.6 (+/- 0.6). There were 98 (65.3%) patients with a score > 1 point at baseline, but only 46 (27.4%) after 3 years and 34 (21.8%) after 5 years. Moreover, 90% of the patients had an improvement of the disability score. Final HAQ disability was associated with baseline values of HAQ score, Pain, Ritchie index, tender joint count, Disease Activity Score, erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), and erosion. Multivariate analysis selected baseline HAQ score, Ritchie index, ESR, CRP, and presence of erosion as independent prognostic factors of HAQ disability. The probability cutoff in the logistic model was selected to minimize the sum of false positive and false negative values: negative predictive value = 92.71%, positive predictive value = 46.15%, p = 0.408. Sex, age, IgM and IgA rheumatoid factors, other tested autoantibodies, and HLA class II genes did not contribute significantly to prediction of the disability after 5 years. At baseline, mean scores were 3.6 units (+/- 7.7) for total radiological score, 1.7 (+/- 4.5) for erosion score, and 1.9 (+/- 3.7) for joint space narrowing score. After 5 years, they were 17.9 +/- 22.3, 6.9 +/- 9.5, and 11.0 +/- 15.4, respectively. No erosion was present at the start in 58.0% of patients, compared to 24.2% and 22.4% at 3 and 5 years. Global radiographic progression concerned 87 patients (55.8%) during the 5 years. CONCLUSION: During the first 5 years of RA, radiological damage increased progressively in half of the patients, whereas HAQ disability improved in most of them during the same period of time and could be predicted by baseline values of HAQ score, Ritchie index, ESR, CRP, and presence (or absence) of erosion.

Adult↗
PubMed2006

A smoking cessation intervention for the methadone-maintained.

AIM: To test, in combination with the nicotine patch, the incremental efficacy of a maximal, tailored behavioral treatment over a minimal treatment for smoking cessation. DESIGN: Randomized clinical trial with 6-month follow-up. SETTING: Five methadone maintenance treatment centers in Rhode Island. PARTICIPANTS: Three hundred and eighty-three methadone-maintained smokers. INTERVENTION: Participants were assigned randomly to nicotine patch (8-12 weeks) plus either (1) a baseline tailored brief motivational intervention, a quit date behavioral skills counseling session and a relapse prevention follow-up session (Max) or (2) brief advice using the National Cancer Institute's 4 As model (Min). An intent-to-treat analysis with those lost to follow-up assumed to smoke was used. MEASUREMENTS: Carbon monoxide (CO)-confirmed 7-day point smoking cessation prevalence at 3 and 6 months, and self-reported numbers of cigarettes smoked per day. FINDINGS: Participants had a mean age of 40 years, were 53% male, 78% Caucasian, smoked 26.7 (+/- 12.2) cigarettes/day and had a mean methadone dose of 95.5 mg. At 3 months, 317 (83%) were re-interviewed; at 6 months, 312 (82%) were re-interviewed. The intent-to-treat, 7-day point prevalence estimate of cessation was 5.2% in the Max group and 4.7% in the Min group (P=0.81) at 6 months. In logistic models with treatment condition, age, gender, race, Fagerström Test for Nicotine Dependence and cigarettes per day as covariates, males were more likely to be abstinent at 3 months (OR 4.67; P=0.003) and 6 months (OR 4.01; P=0.015). CONCLUSION: A tailored behavioral intervention did not increase quit rates over patch and minimal treatment. Smoking cessation rates in methadone-maintained smokers are low, with men having greater success.

Adult↗
PubMed1999

Biopsychobehavioral model of risk of severe hypoglycemia. Self-management behaviors.

OBJECTIVE: To identify self-management antecedents of low blood glucose (BG) (< 3.9 mmol/l) that might be easily recognized, treated, or avoided altogether. RESEARCH DESIGN AND METHODS: Ninety-three adults with type 1 diabetes (age, 35.8 +/- 8 years [mean +/- SD]; duration of diabetes, 17.0 +/- 11 years; daily insulin dose, 0.58 +/- 0.18 U/kg; and HbAlc, 8.6 +/- 1.8%) were recruited to participate in the study. Of the 93 subjects, 42 had a history of severe hypoglycemia (SH), defined as two or more hypoglycemic episodes in the preceding 12 months, and 51 subjects had no history of SH (No-SH) in the same time period. Before each of 70 BG measurements obtained over a 3-week period, subjects used a handheld computer to record whether their most recent insulin, food, and exercise was more than, less than, or the same as usual. Associations among self-management behaviors preceding BG readings < 3.9 mmol/l versus those preceding BG readings of 5.6-7.8 mmol/l were determined using chi 2 tests, analyses of variance, and logistic regression analyses. RESULTS: Analysis of 6,425 self-management/self-monitoring of BG events revealed that the usual amounts of insulin, food, and exercise preceded the events 58.3% of the time. No significant differences were observed for changes in insulin before readings of BG < 3.9 mmol/l versus 7.8 < BG > 5.6 mmol/l, but significantly less food (P < 0.01) was eaten and more exercise (P < 0.001) was performed before the low BG measurement. No interactions between SH and No-SH groups and management behaviors were observed. However, each of the three management variables entered significantly in a logistic model that predicted 61% of all readings of BG < 3.9 mmol/l. CONCLUSIONS: Subjects with a history of SH did not report managing their diabetes differently from those with no such history. Specifically, when low BG occurred, the preceding management behaviors, although predictive of low BG, were not different in SH and No-SH subjects. Overall, self-management behaviors did not distinguish SH from No-SH subjects. Thus, even though it might be beneficial for all patients to review their food and exercise management decisions to reduce their frequency of low BG, an educational intervention whose content stresses insulin, food, and exercise would be unlikely by itself to be sufficient to reduce the frequency of SH.

Adult↗
PubMed2004

Modelling and predicting the simultaneous growth of Listeria monocytogenes and spoilage micro-organisms in cold-smoked salmon.

AIMS: To evaluate and model the simultaneous growth of Listeria monocytogenes and spoilage micro-organisms in cold-smoked salmon. METHODS AND RESULTS: Growth kinetics of L. monocytogenes, lactic acid bacteria (LAB), Enterobacteriaceae, enterococci and Photobacterium phosphoreum were determined in two series of challenge tests with sliced and vacuum-packed cold-smoked salmon (SVP-CSS). The product contained a high level of smoke components and at 2 degrees C levels of L. monocytogenes increased <100-fold in 193 days. Without the addition of spoilage micro-organisms, L. monocytogenes reached ca 108 CFU g-1 at 5, 10, 17.5 and 25 degrees C. Inoculation with spoilage micro-organisms reduced this level to 102-104 CFU g-1. LAB dominated the spoilage microfora of SVP-CSS and competition between LAB and L. monocytogenes in SVP-CSS was appropriately described by a simple expansion of the Logistic model. This interaction model aided in predicting the growth of L. monocytogenes in naturally contaminated SVP-CSS when it was used in combination with expanded versions of existing secondary models for L. monocytogenes and LAB. CONCLUSIONS: Temperature, water activity/NaCl, simultaneous growth of LAB, smoke components and to a lesser extent lactate and pH control growth of L. monocytogenes in SVP-CSS. These factors must be included in mathematical models to predict growth of the pathogen in this product. SIGNIFICANCE AND IMPACT OF THE STUDY: The suggested predictive model can be used to support assessment and management of the human health risk due to L. monocytogenes in SVP-CSS.

Animals↗
PubMed2003

Genetic Analysis Workshop 13: simulated longitudinal data on families for a system of oligogenic traits.

The Genetic Analysis Workshop 13 simulated data aimed to mimic the major features of the real Framingham Heart Study data that formed Problem 1, but under a known inheritance model and with 100 replicates, so as to allow evaluation of the statistical properties of various methods. The pedigrees used were the 330 real pedigree structures (comprising 4692 individuals) with some minor changes to protect confidentiality. Fifty trait genes and 399 microsatellite markers were simulated by gene dropping on 22 autosomal chromosomes. Assuming random ascertainment of families, a system of eight longitudinal quantitative traits (designed to be similar to those in the real data) was generated with a wide range of heritabilities, including some pleiotropic and interactive effects. Genes could affect either the baseline level or the rate of change of the phenotype. Hypertension diagnosis and treatment were simulated with treatment availability, compliance, and efficacy depending on calendar year. Nongenetic traits of smoking and alcohol were generated as covariates for other traits. Death was simulated as a hazard rate depending upon age, sex, smoking, cholesterol, and systolic blood pressure. After the complete data were simulated, missing data indicators were generated based on logistic models fitted to the real data, involving the subject's history of previous missing values, together with that of their spouses, parents, siblings, and offspring, as well as marital status, only-child indicators, current value at certain simulated traits, and the data collection pattern on the cohort into which each subject was ascertained.

Adult↗
PubMed2005

High-level transient expression of recombinant protein in lettuce.

Transient expression following agroinfiltration of plant tissue was investigated as a system for producing recombinant protein. As a model system, Agrobacterium tumefaciens containing the beta-glucuronidase (GUS) gene was vacuum infiltrated into lettuce leaf disks. Infiltration with a suspension of 10(9) colony forming units/mL followed by incubation for 72 h at 22 degrees C in continuous darkness produced a maximum of 0.16% GUS protein based on dry tissue or 1.1% GUS protein based on total soluble protein. This compares favorably to expression levels for commercially manufactured GUS protein from transgenic corn seeds. A. tumefaciens culture medium pH between 5.6 and 7.0 and surfactant concentrations < or = 100 ppm in the vacuum infiltration did not affect GUS expression, while infiltration with an A. tumefaciens density of 10(7) and 10(8) colony forming units/mL, incubation at 29 degrees C, and a surfactant concentration of 1,000 ppm significantly reduced expression. Incubation in continuous light caused lettuce to produce GUS protein more rapidly, but final levels did not exceed the GUS production in leaves incubated in continuous darkness after 72 h at 22 degrees C. The kinetics of GUS expression during incubation in continuous light and dark were represented well using a logistic model, with rate constants of 0.30 and 0.29/h, respectively. To semi-quantitatively measure the GUS expression in large numbers of leaf disks, a photometric enhancement of the standard histochemical staining method was developed. A linear relationship with an R2 value of 0.90 was determined between log10 (% leaf darkness) versus log10 (GUS activity). Although variability in expression level was observed, agroinfiltration appears to be a promising technology that could potentially be scaled up to produce high-value recombinant proteins in planta.

Agrobacterium tumefaciens↗
PubMed2006

Anxiety in persons 75 and older: findings from a tri-ethnic population.

PURPOSE: Little research has been reported about anxiety in older populations. We assessed the prevalence of anxiety and examined associations between anxiety and sociodemographic, physical, mental, and functional health characteristics in an older tri-ethnic population. DESIGN AND METHODS: A cross-sectional, population-based study of older, noninstitutionalized non-Hispanic Whites, non-Hispanic Blacks, and Hispanics was conducted from the baseline assessment of the Health of the Public (HoP) database. Measurements included a self-report anxiety scale and physical, mental, and functional health. Data were analyzed with general linear models and logistic regression models. RESULTS: Overall, 31.4% reported anxiety (score > or = 45); 32.3% of woman and 30.5% of men (P=.67). Older Hispanics (22.2%) reported the lowest prevalence of anxiety followed by non-Hispanic Blacks (26.6%) and non-Hispanic Whites (44.3%) (P=.0001). Common predictors of anxiety seen after linear and logistic regression included being married, White, and increasing number of medications and depressive symptoms. IMPLICATIONS: Anxiety is prevalent in older adults. The findings also indicate higher rates of anxiety in older non-Hispanic Whites compared with older non-Hispanic Blacks and Hispanics.

Aged↗
PubMed1998

Multivariate analysis of factors associated with postoperative pulmonary complications following general elective surgery.

OBJECTIVE: To develop a predictive model identifying perioperative conditions associated with postoperative pulmonary complications (PPCs). DESIGN: A prospective survey of patients whose preoperative history and physical examination, spirometric, PaO2 and PaCO2 analysis, and operative results were recorded. These patients underwent postoperative cardiopulmonary examinations until they were discharged from the hospital; their medical records were also reviewed until they were discharged from the hospital. SETTING: The Louisville Veterans Administration Medical Center, Louisville, Ky. PATIENTS: A randomly chosen sample of patients aged 40 years or older who required elective, nonthoracic surgery under general or spinal anesthesia and who were hospitalized at least 24 hours postoperatively. MAIN OUTCOME MEASURE: An analysis of risk factors associated with the development of 1 or more of the following conditions: acute bronchitis, bronchospasm, atelectasis, pneumonia, adult respiratory distress syndrome, pleural effusion, pneumothorax, prolonged mechanical ventilation, or death secondary to acute respiratory failure. RESULTS: Postoperative pulmonary complications developed in 16 (11%) of 148 patients. The risk factors found to be higher among those with PPCs compared with those without PPCs were postoperative nasogastric intubation (81% vs 16%, P<.001), preoperative sputum production (56% vs 21%, P=.005), and longer anesthesia duration (480 vs 309 minutes, P<.001). Upper abdominal surgery was performed in 11 (69%) of the 16 patients with PPCs and in 20 (15%) of the 132 patients without PPCs (P<.001); this difference lost significance in multivariate analysis. The final linear logistic model included postoperative nasogastric intubation (odds ratio [OR], 21.8), preoperative sputum production (OR, 4.6), and longer anesthesia duration (OR exp[0.01x] for an increase in x minutes) (1 minute of additional anesthesia time increases the OR to 1.01), resulting in 92% accuracy in predicting PPCs. CONCLUSIONS: We identified 3 potentially modifiable risk factors for PPCs. If validated, our results may lead to modifications of perioperative care that will further reduce PPCs.

Adult↗
PubMed1994

Artificial intelligence versus logistic regression statistical modelling to predict cardiac complications after noncardiac surgery.

The traditional approach to developing models predictive of cardiac events has been to perform logistic regression (LR) analysis on a variety of potential predictors. An alternative to use an artificial intelligence system called a neural network (NN) which simulates biological intelligence. To evaluate the potential applicability of the latter method, we compared the ability of LR and NN techniques to predict cardiac events after noncardiac surgery. A total of 200 patients (training group) underwent cardiac risk assessment before major noncardiac surgery using 17 clinical parameters and 7 quantitative indices based on dipyridamole-thallium imaging. There were 21 post-operative myocardial infarctions and/or cardiac deaths. Data from the training group were used to develop two predictive models: one based on backward stepwise LR multivariate statistical analysis and the other one using a neural network. Both models were then validated on a second group of 160 consecutive patients also referred for preoperative risk stratification (validation group). The NN consisted of 14 input, 29 hidden, and 1 output neurons and used a back-propagation algorithm (learning rate 0.2, training tolerance 0.5, sigmoid transfer function). The sensitivity, specificity, positive and negative predictive accuracies for the prediction of postoperative events in the validation group of 160 patients were, respectively, 67% (6/9), 82% (124/151), 18% (6/33), and 98% (124/127) for LR, and 67% (6/9), 96% (145/151), 50% (6/12), and 98% (145/148) for the NN, with a difference in specificity which attained statistical significance (p < 0.01). Artificial intelligence may provide a useful alternative to conventional LR statistical analysis for the purpose of preoperative cardiac risk assessment.

Artificial Intelligence↗
PubMed2005

Extending logistic regression to model diffuse interactions.

In an observational study focussed on association between a health outcome and numerous explanatory variables, the question of interactions can be problematic. Commonly, logistic regression of the outcome on the explanatory variables might be employed. Such modelling often includes an attempt to select some pairwise product interaction terms, from amongst the many such possible pairs. For several reasons, however, this can be unsatisfying. Here we consider a different approach based on a parsimonious extension of a logistic regression model without interaction terms. This extension permits an overall synergism or antagonism in how the explanatory variables combine to associate with the outcome, without any attempt to identify specific variables which give rise to interactive behaviour. We call this diffuse interaction. We elucidate some simple properties of the diffuse interaction model, and give an example of its application to epidemiological data. We also consider asymptotic behaviour in a restricted case of the model, to gain some insight into how well this kind of interaction can be detected from data.

Analysis of Variance↗
PubMed2005

A logistic regression mixture model for interval mapping of genetic trait loci affecting binary phenotypes.

Often in genetic research, presence or absence of a disease is affected by not only the trait locus genotypes but also some covariates. The finite logistic regression mixture models and the methods under the models are developed for detection of a binary trait locus (BTL) through an interval-mapping procedure. The maximum-likelihood estimates (MLEs) of the logistic regression parameters are asymptotically unbiased. The null asymptotic distributions of the likelihood-ratio test (LRT) statistics for detection of a BTL are found to be given by the supremum of a chi2-process. The limiting null distributions are free of the null model parameters and are determined explicitly through only four (backcross case) or nine (intercross case) independent standard normal random variables. Therefore a threshold for detecting a BTL in a flanking marker interval can be approximated easily by using a Monte Carlo method. It is pointed out that use of a threshold incorrectly determined by reading off a chi2-probability table can result in an excessive false BTL detection rate much more severely than many researchers might anticipate. Simulation results show that the BTL detection procedures based on the thresholds determined by the limiting distributions perform quite well when the sample sizes are moderately large.

Algorithms↗