Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Discriminant Analysis”

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 181 records · Page 10Linked to original sources
PubMed1987

A comparison of discriminant analysis and logistic regression for the prediction of coliform mastitis in dairy cows.

Results from discriminant analysis and logistic regression were compared using two data sets from a study on predictors of coliform mastitis in dairy cows. Both techniques selected the same set of variables as important predictors and were of nearly equal value in classifying cows as having, or not having mastitis. The logistic regression model made fewer classification errors. The magnitudes of the effects were considerably different for some variables. Given the failure to meet the underlying assumptions of discriminant analysis, the coefficients from logistic regression are preferable.

Animals↗
PubMed1981

Early prediction of symptomatic patent ductus arteriosus from perinatal risk factors: a discriminant analysis model.

A scoring system based on discriminant analysis was devised to predict, within 24 hours after birth, whether or not a premature infant will subsequently develop a symptomatic PDA. Five risk factors including birth weight, the diagnosis of hyaline membrane disease, intrauterine growth retardation, acute perinatal stress, and treatment with distending airway pressure were reduced to a discriminant score which separated infants with symptomatic PDA from infants without symptomatic PDA. Based on this score, the likelihood that an individual infant would later develop symptomatic PDA could be expressed as a probability function. When applied prospectively, this score predicted the correct outcome of 80% of infants in a test population. This predictive model should be useful in clinical trials and other applications requiring a quantitative expression of risk for developing symptomatic ductus shunting.

Birth Weight↗
PubMed1986

Electrocardiographic diagnosis of posterior myocardial infarction revisited: a new approach using a multivariate discriminant analysis and thallium-201 myocardial scintigraphy.

This study examined the feasibility of using a multivariate discriminant analysis to design a useful electrocardiographic (ECG) model to diagnose posterior myocardial infarction (MI). Thallium-20) scintigraphy was used as a reference standard to identify posterior scar (fixed perfusion defects). The model was derived from 111 patients of whom 37 had fixed posterior defects and 74 had normal images, and its validity was subsequently tested in a separate group of 180 patients. In the initial group of patients, the fixed perfusion defects involved the posterior left ventricular wall alone in 15 patients, and the posterior and inferior walls in 22 patients. Stepwise multivariate discriminant analysis of 26 ECG variables produced a model of two variables (Q-wave duration in a VF and T-wave amplitude in V1) which provided a sensitivity of 78%, a specificity of 89%, and a predictive accuracy of 86% for the diagnosis of posterior MI. This model, when tested in the second group of 180 patients, yielded an overall prediction accuracy of 82% (sensitivity 65%, specificity 85%). Thus, the combination of Q-wave in a VF and upright T wave in V1 is the best ECG predictor of posterior MI. These two variables reflect the frequent association of posterior MI with inferior MI, and the reciprocal repolarization changes in the right precordial leads.

Adult↗
PubMedDate not supplied

Classification of passive auditory event-related potentials using discriminant analysis and self-organizing feature maps.

Discriminant analysis (DA) and self-organizing feature maps (SOFM) were used to classify passively evoked auditory event-related potentials (ERP) P(1), N(1), P(2) and N(2). Responses from 16 children with severe behavioral auditory perception deficits, 16 children with marked behavioral auditory perception deficits, and 14 controls were examined. Eighteen ERP amplitude parameters were selected for examination of statistical differences between the groups. Different DA methods and SOFM configurations were trained to the values. SOFM had better classification results than DA methods. Subsequently, measures on another 37 subjects that were unknown for the trained SOFM were used to test the reliability of the system. With 10-dimensional vectors, reliable classifications were obtained that matched behavioral auditory perception deficits in 96%, implying central auditory processing disorder (CAPD). The results also support the assumption that CAPD includes a 'non-peripheral' auditory processing deficit.

Adolescent↗
PubMed1986

On robust partial discriminant analysis as a decision-making tool with clinical and analytical chemical data.

Classification is one of the fundamental goals of science and is basic to the diagnosis of disease. Unfortunately, classifying objects (e.g., patients) on the basis of clinical and/or laboratory experimental observations into various groups can be difficult when the groups overlap or contain outlying points. Recently, Broffitt, Randles, and co-workers proposed a procedure, robust partial discriminant analysis (RPDA) for dealing with such problems, but testing of the procedure was limited to Monte Carlo simulation. In this study, RPDA was applied to real data, in order to compare its effectiveness with ordinary discriminant analysis, as well as to determine if RPDA was a suitable procedure to use to classify chemical compounds on the basis of experimental observations and as a tool in the diagnosis of disease (in particular, multiple sclerosis and thyrotoxicosis), with data based on experimental and clinical observations. The resulting RPDA classifications were an improvement over those obtained from ordinary discriminant analysis.

Aldehydes↗
PubMed1995

An investigation of the relationship between antioxidant vitamin intake and coronary heart disease in men and women using discriminant analysis.

Smoking, high blood pressure and elevated blood cholesterol are the well-established 'classical' risk factors for coronary heart disease (CHD) in men and women. However, it is also well-known that there is a considerable degree of residual variation in CHD after these factors have been taken into account. Consideration of antioxidant vitamin status may help to reduce this unexplained variation. Here, discriminant analysis is applied to the baseline cross-sectional data from the Scottish Heart Health Study. The problem of possible behavioural changes after diagnosis for CHD is addressed by analysing diagnosed and undiagnosed CHD cases separately. Results show that the combined dietary intakes of the antioxidant vitamins C, E and carotene (assessed using a food frequency questionnaire) differentiate CHD prevalence as well as do the classical risk factors. For women, stepwise discriminant analysis shows that the effect of the antioxidant vitamins on CHD is removed by adjustment for the classical risk factors and age. For men, however, the antioxidant vitamins still contribute to the discriminant function. It is concluded that dietary antioxidant vitamins appear to have a significant effect on the prevalence of CHD, especially amongst men. The benefits and problems of using discriminant analysis in this practical context are discussed, including the assumptions that need to be tested.

Adult↗
PubMed1997

[Clinical and biological markers of secondary uveitis: results of a discriminant analysis].

BACKGROUND: The association of uveitis and systemic disease is well known. Patients suffering from uveitis often undergo a extensive battery of tests in order to detect underlying disease, but the efficiency of such screening is uncertain. The aim of this study was to investigate useful clinical data for recognizing secondary uveitis. PATIENTS AND METHODS: We conducted a prospective analysis of 115 patients with uveitis of unknown etiology. All of them were included in an extensive protocol study. Four groups were considered: specific ocular disease (SOD), idiopathic uveitis, HLA-B27 associated uveitis without arthritis (HLA-B27-AU) and secondary uveitis. Groups were compared by analysis of variance for continuous variables, and chi 2 test or Student's t-test for discrete variables. A stepwise multiple discriminant analysis was performed for ranking the variables in order of their usefulness for distinguishing idiopathic and secondary uveitis. RESULTS: We diagnosed 11 SOD (9.6%), 54 idiopathic uveitis (47%), 6 HLA-B27-AU (5.2%) and 41 secondary uveitis (35.7%). The discriminant analysis showed that age, an elevated erythrocyte sedimentation rate, presence of cutaneous lesion, joint pain and genital ulcers are the strongest predictors of secondary uveitis. This model classification functions detected 92.5% of idiopathic uveitis and 72% of secondary uveitis. The global percentage of patients with a correct diagnosis was 84.6%. CONCLUSIONS: Anamnesis, physical examination and basic laboratory tests are sufficient tools for the diagnostic approach of the majority of patients with uveitis. Subsequent diagnostic procedures must be planned in each patient to confirm a specific disease.

Adolescent↗
PubMed1975

A simulation study of the efficacy of stepwise discriminant analysis in the detection and comparison of event related potentials.

Cortical average evoked potentials were simulated by summing five damped sinusoids. The characteristics of these "evoked" responses could be manipulated by changing parameters of the sinusoids. The synthesized signals were mixed with noise processes whose power and band-width were manipulated. Thus data were generated to stimulate a variety of conditions which could conceivably occur in an experiment on evoked potentials. Stepwise discriminant analysis (BMD07M) has been applied to these simulated data in an attempt to determine the degree to which the program identifies, in a sensible manner, the differences we introduced into the synthesized evoked responses. The simulation results indicate that stepwise discriminant analysis can indeed be an efficacious tool in research on evoked potentials. The program does detect differences in evoked potentials. It can be used, with some reservations, to identify the components of an evoked potential which the experimental variables have affected. In a special set of simulations we have attempted to determine the degree to which stepwise discriminant analysis could serve to detect the presence or absence of an evoked potential. These simulations show that the score of an average evoked potential in the data. The implications of this finding to the use of evoked potentials in sensory sensitivity testing were evaluated in studies for the effect on them of stimulus intensity.

Electroencephalography↗
PubMed1983

The use of multivariate discriminant analysis in the antenatal detection of fetal neural tube defects.

Early second trimester amniotic fluid is classified by stepwise discriminant analysis involving the computer program BMDP77, based upon biochemical analyses for alpha-fetoprotein and eight other easily assayed variables. The outcome of about 800 pregnancies has been predicted on this basis, over 80 of which were associated with fetal neural tube defects. Classification is shown to be more reliable than using alpha-fetoprotein and gestational age alone. Although in practice this approach has now been superceded by the introduction of the analysis of acetylcholine-esterase isoenzyme activity in amniotic fluid, the results demonstrate a successful application of discriminant analysis to medical diagnosis.

Amniocentesis↗
PubMed2003

Coupling proton transfer reaction-mass spectrometry with linear discriminant analysis: a case study.

Proton transfer reaction-mass spectrometry (PTR-MS) measurements on single intact strawberry fruits were combined with an appropriate data analysis based on compression of spectrometric data followed by class modeling. In a first experiment 8 of 9 different strawberry varieties measured on the third to fourth day after harvest could be successfully distinguished by linear discriminant analysis (LDA) on PTR-MS spectra compressed by discriminant partial least squares (dPLS). In a second experiment two varieties were investigated as to whether different growing conditions (open field, tunnel), location, and/or harvesting time can affect the proposed classification method. Internal cross-validation gives 27 successes of 28 tests for the 9 varieties experiment and 100% for the 2 clones experiment (30 samples). For one clone, present in both experiments, the models developed for one experiment were successfully tested with the homogeneous independent data of the other with success rates of 100% (3 of 3) and 93% (14 of 15), respectively. This is an indication that the proposed combination of PTR-MS with discriminant analysis and class modeling provides a new and valuable tool for product classification in agroindustrial applications.

Discriminant Analysis↗
PubMed2000

Linear discriminant analysis of symptoms in patients with chronic constipation: validation of a new scoring system (KESS).

PURPOSE: The aim of this study was to devise a symptom scoring system to assist in diagnosing constipation and in discriminating among pathophysiologic subgroups. METHOD: A structured symptom scoring questionnaire (11 questions) was completed by 71 chronically constipated patients and by 20 asymptomatic controls. The symptom score was correlated with a previously validated constipation score (Cleveland Clinic Score). All patients underwent colonic transit studies, standard anorectal physiology testing, and evacuation proctography. On the basis of these investigations alone, an observer blinded to the questionnaire results allocated patients to one of three pathophysiologic subgroups: slow-transit constipation, rectal evacuatory disorder, or mixed (slow-transit constipation and rectal evacuatory disorder). Linear discriminant analysis was used to assess the ability of different questionnaire symptoms to discriminate among these subgroups. RESULTS: Total symptom score was strongly correlated with the Cleveland Clinic Score (r = 0.9). The median total score in constipated patients was 20 (range, 11-35) compared with a median of 2 in controls (range, 0-6). Discriminant analysis using cross validation estimated that pathophysiology could be predicted correctly for 55 percent (95 percent confidence interval = 43-67 percent) of patients using just five symptoms. The discriminant function rarely misclassified patients with rectal evacuatory disorder as slow-transit constipation and vice versa, but could not effectively discriminate between patients with single and mixed pathologies. CONCLUSION: This new scoring system is a valid technique to assist in the diagnosis of constipation and is the first study using appropriate statistical methodology to demonstrate a discriminatory ability of multiple symptoms in constipation. At present, symptom analysis does not adequately differentiate major pathophysiologic subgroups for use in clinical practice.

Adult↗
PubMed1988

[Assessment of the discriminant analysis concerning the prognostic factors in patients with differentiated thyroid cancer].

We reported the result of our discriminant analysis concerning prognostic factors in differentiated thyroid cancer, which based on 134 cases obtained from three hospitals. The sensitivity, specificity and false-negative rate were 59%, 61% and 22%, respectively. When these values were estimated individually in each hospital, the figures were better than those obtained from a whole group of patients. The purpose of this paper is to assess the reliability of the previous result. Simulation was done to assess the relationship between the number of patients and the ratio of criterion variables. New groups were made from the 134 patients by random sampling. In each group, discriminant analysis was done by using the same nine explanatory variables as in the previous report (age, sex, diameter of tumor, site of tumor, histology, local invasion, lymph node metastasis, operative method and lymph node dissection). The results revealed that sensitivity was relatively stable but specificity and false-negative rate were better in the smaller group of patients. The ratio of criterion variables had an effect on the results. In the higher ratio, sensitivity was higher and the false-negative rate was lower.

Adenocarcinoma↗
PubMed1980

Multivariate discriminant analysis of the clinical and angiographic predictors of operative mortality from the Collaborative Study in Coronary Artery Surgery (CASS).

The Collaborative Study in Coronary Artery Surgery (CASS) is a large multi-institutional study of the medical and surgical treatment of coronary artery disease (CAD). Fifteen cooperating institutes have carried out isolated coronary artery bypass grafting (CABG) on 6,176 patients from August, 1975, through December, 1978. The operative mortality (OM) was 2.3%. In an effort to better understand the clinical and angiographic characteristics predictive of OM, we have done a multivariate discriminant analysis of variables associated with OM. Numerous clinical and angiographic variables were selected from the CASS data file and evaluated in a univariate manner for their relationship to OM. Twenty of these variables were then selected for multivariate discriminant analysis. Clinical variables of most predictive value were age, female sex, increased heart size, and congestive heart failure (CHF). Angiographic variables of importance included left ventricular wall motion abnormalities, and left main coronary disease (LMCD). The priority of operation (elective, urgent, or emergent) was also associated with OM. Six variables that contained the most predictive information were selected by discriminant analysis for a group of 6,176 patients who had isolated bypass operations. In descending order of importance they were age, left main coronary artery stenosis greater than or equal to 90%, female sex, left ventricular wall motion score, left ventricular end-diastolic pressure (LVEDP), and râles. Five other groups or subgroups of patients were also analyzed in a similar manner. There is a strong association of OM with advanced age, female sex, and variables associated with left ventricular dysfunction. The risk of OM for an individual patient may be estimated with the use of these clinical and angiographic characteristics.

Adult↗
PubMed1994

Nuclear quantitative grading by discriminant analysis of renal cell carcinoma samples. A patient survival evaluation.

Specimens from 60 cases of renal cell carcinoma (RCC) were graded employing quantitative nuclear data combined with multivariate discriminant analysis. Evaluation of patient survival was analysed with respect to quantitative microscopic and qualitative features. Both morphometric and stereological estimators were used to establish the nuclear size and form pattern of the RCC specimens. Tumoural dedifferentiation paralleled progressive increases in nuclear elongation and in two- and, especially, three-dimensional--mean nuclear volume (MNV)--size parameters. Using stepwise discriminant analysis, 85.0 per cent of the specimens were correctly classified when differentiating grade 2 and 3 tumours. It is concluded that simple and realistic estimates of MNV are the best discriminator for objective grading in patients with RCC. Univariate survival analysis demonstrated the important significance of several features such as MNV, clinical stage, and nuclear discriminant and histopathological tumour grades. Nuclear form factor PE, area, and perimeter were also significant. A prognosis study based on the Cox model using a stepwise selection of parameters showed that only MNV has an independent prognostic role when examining all investigated quantitative parameters. The clinical stage was the best prognostic feature when all quantitative and qualitative characteristics were included in the analysis.

Adult↗
PubMed1998

Screening for primary aldosteronism with a logistic multivariate discriminant analysis.

OBJECTIVE: Primary aldosteronism (PA) is the most common endocrine cause of curable hypertension, but no single test unequivocally identifies it. Accordingly, we investigated the usefulness of a logistic multivariate discriminant analysis (MDA) approach for PA screening. DESIGN: Generation of a logistic MDA function based on retrospective analysis of biochemical tests in a large cohort of referred patients with/without confirmed Conn's adenoma (CA), followed by prospective validation of the model. PATIENTS: We investigated 574 selected hypertensives: 206 (32 with and 174 without CA) retrospectively, 48 (with a 13% prevalence of CA) prospectively for the validation of the model, and 320 referred hypertensives (with a 3.4% prevalence of CA) similarly evaluated. Patients were referred to a specialised centre for hypertension (4th Clinica Medica--University of Padua) and to a department of Internal Medicine of a regional hospital (Reggio Emilia). MEASUREMENTS: In all patients we measured several demographic and biochemical variables and performed a captopril test. A stepwise analysis of variance, based on a model fitted with several different variables, identified baseline (sALDO) and captopril-suppressed plasma aldosterone (cALDO), supine plasma renin activity (sPRA) and K+ as the most informative. Therefore, two models of logistic MDA with sPRA, K+, and either sALDO (model A) or cALDO (model B) were developed and used. ROC analysis was also performed to assess the optimal cut-off values. RESULTS: The model B of MDA provided the best performance and identified CA with 100% sensitivity and 81% accuracy. When used prospectively it showed 100% sensitivity, both in the Padua (88% accuracy) and in the Reggio Emilia series (90% accuracy). However, at both institutions most patients with idiopathic hyperaldosteronism (IHA) were also detected. CONCLUSIONS: Thus, although developed from patients with confirmed Conn's adenoma, a strategy based on multivariate discriminant analysis can be used prospectively for accurate screening for primary aldosteronism. Furthermore, it was proven to be accurate and applicable to patients tested with similar modalities at a different institution. Although this approach did not provide a clear-cut discrimination of Conn's adenoma from idiopathic hyperaldosteronism, it may avoid unnecessary and costly further testing in patients with a low probability of primary aldosteronism.

Adult↗
PubMed1996

Separation of pulmonary disorders with two-dimensional discriminant analysis of crackles.

Previous studies have indicated that disorders producing crackling lung sounds may be different in terms of the waveform of the crackles or their timing in a respiratory cycle. In this study, we evaluated whether two-dimensional discriminant analysis of crackles has a better ability to separate pulmonary disorders than does a single-dimensional analysis. Cracking sounds of patients with cryptogenic fibrosing alveolitis (n = 10), bronchiectasis (n = 10), COPD (n = 10), heart failure (n = 10) and acute pneumonia (n = 11) and of those recovering from pneumonia (n = 9) have been studied. Variables indicating the timing of crackles during inspiration (beginning and endpoint of crackling) and their waveform (initial deflection width (IDW), two cycle duration (2CD) and largest deflection width (LDW)), were used for the analysis. The discrimination properties of one- and two-dimensional analyses with these variables were compared. The two-dimensional distances between the patient groups were the largest by combining IDW and the end-point of crackling. Cryptogenic fibrosing alveolitis was distinguished from bronchiectasis, COPD, heart failure and acute pneumonia without overlap. The differences between the diseases were illustrated two-dimensionally with ellipses. The two-dimensional analysis resulted in better separation between the groups than the use of single characteristics alone. This type of analysis can enhance the diagnostic power of acoustic pulmonary studies. It is also an informative visual way to find differences among pulmonary disorders.

Acute Disease↗
PubMed1993

Differences in BAL fluid variables in interstitial lung diseases evaluated by discriminant analysis.

The aim of this study was to investigate the possibility of distinguishing between patients with similarities in clinical presentation, suffering from three frequently occurring interstitial lung diseases, by means of discriminant analysis, using a number of selected variables derived from bronchoalveolar lavage fluid (BALF) analysis. The study involved all 277 patients, who had an initial bronchoalveolar lavage (BAL) in the period 1980-1990. These patients belonged to the following diagnostic groups: sarcoidosis (n = 193), subacute extrinsic allergic alveolitis (EAA) (n = 39) and idiopathic pulmonary fibrosis (IPF) (n = 45). Thirty healthy volunteers were used as controls. Cellular and non-cellular constituents of BALF were evaluated. Variables, which could be used to discriminate among the three diagnostic groups were: yield of recovered BALF, total cell count, and percentages of alveolar macrophages, lymphocytes, polymorphonuclear neutrophils, eosinophils and plasma cells in BALF. When the set of data used to predict the membership of patients to diagnostic groups (test set) was the same as that in which the discriminant analysis was performed (learning set), 93% of the cases were correctly classified. This percentage decreased to 90%, however, when the test set was different from the learning set. It is possible to discriminate among patients with sarcoidosis, EAA or IPF with these selected variables. It appears that bronchoalveolar lavage (BAL) is useful as an adjunct in concert with other diagnostic methods.

Adolescent↗
PubMed1983

Discriminant analysis of data in enzyme immunoassay.

A critical step in the development of both qualitative and quantitative enzyme immunoassays is establishing the positive/negative discrimination, or cut-off, value. Data derived from an indirect immunofluorescence assay, hemagglutination inhibition, and enzyme immunoassay to detect IgG antibodies to measles virus were applied to a discriminant analysis program to determine the positive/negative cut-off value. Application of the discriminant analysis demonstrated a greater utilization of the sensitivity of the enzyme immunoassay than the most commonly used methods. This method also illustrates the importance of examining both antibody positive and negative sera, rather than negative sera alone, in determining the cut-off value. In addition, probability of membership in the antibody positive or negative group is included in the determination. This increases the information base for risk assessment and clinical evaluation.

Antibodies, Viral↗