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 271 records · Page 15Linked to original sources
PubMed1970

The use of discriminant analysis for examining the histological feathers of oral keratoses and lichen planus.

In a retrospective survey of 235 cases in which the diagnosis on biopsy was lichen planus, keratosis or leukoplakia, the histological features were re-assessed as objectively as possible and without reference to the original diagnosis.The tissue changes were recorded under 39 headings, and many were assessed on a roughly quantitative basis. In addition, two clinical features were included; whether the biopsy was from the buccal mucosa (as opposed to some other intraoral site) and whether the lesions involved multiple intraoral sites. For each possible pair of diagnostic categories (keratosis and leukoplakia, lichen planus and keratosis, lichen planus and leukoplakia) the recorded findings were subjected to discriminant analysis in order to provide a quantitative assessment of the value of each individual feature for discriminating between the two diagnostic groups. The computer programme also provided for the application of these calculated values to yield a "score" for each case, and for an assessment of the significance of the separation of the diagnostic groups thus achieved. In general, the values calculated by the computer for the discriminatory value of each tissue change accorded with our subjective impressions, but a number of features that were given a relatively high value had not previously been recognized as important in differential diagnosisA discriminant analysis was also performed on those cases of leukoplakia known to have later developed a carcinoma, in comparison with the leukoplakia cases that did not develop carcinoma. High values were accorded mainly to the well-known features of epithelial atypia, but a similar high value was indicated for the presence of Russell bodies. we had not previously realized that the presence of Russell bodies was of prognostic significance in this context.When the total scores for the groups of cases were analysed, it was found that the separation of each pair of diagnostic groups was significant at the 1% level. The separation of leukoplakia cases that subsequently developed carcinoma, from those that did not develop carcinoma, was significant at the 5% level. In this latter analysis, a better separation might be achieved with larger numbers of cases, but there will always be the complicating factor that an unknown number of leukoplakia cases would develop carcinoma if the patient had received no treatment.

Biopsy↗
PubMed1998

Differentiation between transmural perioperative myocardial infarction and subendocardial injury after coronary artery bypass grafting using biochemical tests, elaborated by cluster and discriminant analysis.

The aim of this study is to differentiate between transmural perioperative myocardial infarction (T-PMI) and subendocardial perioperative myocardial injury (S-PMI) as a complication of coronary artery bypass grafting (CABG). Seventy-three patients undergoing CABG were followed post operatively by measuring troponin T, CK-MB isoenzyme mass concentration (CK-MB mass), creatine kinase MB isoenzyme activity (CK-MB activity), creatine kinase (CK), alpha hydroxybutyrate dehydrogenase (HBD), and aspartate aminotransferase (AST) at five sampling times. Lacking a proper definition of the gold standard for the diagnosis of perioperative myocardial infarction, a statistical procedure was used. Supported by the cluster analysis method of Ward, patients were assigned to a patient group with a perioperative myocardial infarction (PMI) or a patient group without a PMI (non-PMI) as a confirmation of interpretation of the biochemical results. Using the results of electrocardiogram (ECG) and echocardiography, the PMI patient group was split into a T-PMI patient group and a S-PMI patient group. With discriminant analysis, two canonic discriminant functions were drawn up to differentiate between patients suffering from a T-PMI or S-PMI and non-PMI patients.

Aspartate Aminotransferases↗
PubMed2005

Discriminative analysis of brain function at resting-state for attention-deficit/hyperactivity disorder.

In this work, a discriminative model of attention deficit hyperactivity disorder (ADHD) is presented on the basis of multivariate pattern classification and functional magnetic resonance imaging (fMRI). This model consists of two parts, a classifier and an intuitive representation of discriminative pattern of brain function between patients and normal controls. Regional homogeneity (ReHo), a measure of brain function at resting-state, is used here as a feature of classification. Fisher discriminative analysis (FDA) is performed on the features of training samples and a linear classifier is generated. Our initial experimental results show a successful classification rate of 85%, using leave-one-out cross validation. The classifier is also compared with linear support vector machine (SVM) and Batch Perceptron. Our classifier outperforms the alternatives significantly. Fisher brain, the optimal projective-direction vector in FDA, is used to represent the discriminative pattern. Some abnormal brain regions identified by Fisher brain, like prefrontal cortex and anterior cingulate cortex, are well consistent with that reported in neuroimaging studies on ADHD. Moreover, some less reported but highly discriminative regions are also identified. We conclude that the discriminative model has potential ability to improve current diagnosis and treatment evaluation of ADHD.

Algorithms↗
PubMed1989

Prognostic assessment in stage I ovarian cancer using a discriminant analysis with clinicopathological and DNA flow cytometric data.

We have previously defined three types of tumor DNA histograms, which are associated with favourable, intermediate and poor prognosis of patients with ovarian cancer. In the present study we evaluated the value of DNA histogram type combined with clinicopathological data in predicting long-term clinical outcome in stage I ovarian cancer. A stepwise discriminant analysis was done to find out the best combination of prognostic parameters in the distinction of two groups of stage I ovarian cancer patients; those with less than 5-year overall survival (22 cases) and those with longer than 5-year recurrence-free survival (47 cases). DNA histogram and histological type as well as FIGO stage in that order proved to be the most discriminating parameters and their combination allowed the correct prediction of clinical outcome in 78% of stage I ovarian cancer patients. In stage Ia the proportion of correctly classified cases based on DNA histogram and histological type was 82%. If DNA histogram was omitted from the discriminant analysis, the combination stage and histological type correctly classified a significantly lower percentage (68%) of patients. DNA flow cytometry thus improved the prognostic evaluation in stage I ovarian cancer, but even when combined with conventional clinicopathological factors failed to give correct prognostic assessment in about 20% of patients with stage I ovarian cancer.

Aged↗
PubMed1978

Discriminant analysis of EEG in children with situational, learning and impulse disorders.

Neurological and EEG findings were studied in 113 children diagnosed as adjustment reaction, behavior disorder, and minimal brain dysfunction (MBD). The MBD patients were subdivided into MBD behavior and MBD learning groups, thus yielding four patient groups in all. The presence of soft neurological signs and abnormal EEG examinations increased in stepwise fashion from the adjustment reaction group to the behavior disorder group, to the MBD behavior group, to the MBD learning group. The EEG records were rated on 16 variables, including amounts and frequencies of a variety of normal EEG features as well as abnormal features. The EEG ratings were then subject to a discriminant analysis. Three discriminant functions were found to be necessary to describe the relation of the four groups. The functions were identified as cortical maturity, cortical abnormality, and cortical reactivity. This implies that the groups do not just differ along one dimension, but along three dimensions. The discriminant program correctly classified 53% of the patients.

Attention Deficit Disorder with Hyperactivity↗
PubMed1987

Multivariate discriminant analysis of risk factors for operative mortality following isolated coronary artery bypass graft. Loyola University Medical Center experience, 1970 to 1984.

The Loyola Open-Heart Registry is a fully operational database that contains detailed data on approximately 9,000 patients who have undergone coronary bypass or cardiac valve replacement from January 1970 to December 1984. We analyzed the registry data using multivariate discriminant analysis to identify and quantitate those factors that might predict operative mortality (OM) for patients undergoing coronary artery bypass grafts at Loyola University Medical Center: Operative mortality was defined as death within 30 days following surgery. A total of 50 clinical and angiographic variables were analyzed for possible univariate association with operative mortality. Twenty-two variables were found to have significant univariate association with OM, and these 22 variables were subjected to multivariate discriminant analysis. For patients undergoing isolated, elective coronary artery bypass, the factors found to be predictive of OM are age (greater than 70) (F = 11.57), severe (more than six stenoses) coronary artery disease (F = 5.81), diffuse disease (F = 5.54), positive family history (F = 5.17), and number of coronary arteries bypassed (F = 4.78).

Coronary Artery Bypass↗
PubMed2003

Data reduction using a discrete wavelet transform in discriminant analysis of very high dimensionality data.

We present a method of data reduction using a wavelet transform in discriminant analysis when the number of variables is much greater than the number of observations. The method is illustrated with a prostate cancer study, where the sample size is 248, and the number of variables is 48,538 (generated using the ProteinChip technology). Using a discrete wavelet transform, the 48,538 data points are represented by 1271 wavelet coefficients. Information criteria identified 11 of the 1271 wavelet coefficients with the highest discriminatory power. The linear classifier with the 11 wavelet coefficients detected prostate cancer in a separate test set with a sensitivity of 97% and specificity of 100%.

Data Compression↗
PubMed1981

Identification of patients with cholesterol or pigment gallstones by discriminant analysis of radiographic features.

In a search for a way to distinguish cholesterol gallstones from pigment gallstones by oral cholecystography, we evaluated 56 patients with surgically confirmed cholelithiasis. Only buoyancy was highly predictive of gallstone composition: all 14 patients with floating stones had cholesterol stones (P less than 0.01), but only one third of the patients with cholesterol stones had stone buoyancy. Using a function derived by stepwise discriminant analysis, we separated patients with cholesterol stones from those with pigment stones. The predictive accuracy was significantly improved: sensitivity was 95 per cent (37 of 39 patients with cholesterol stones), specificity was 82 per cent (14 of 17 patients with pigment stones), and efficiency was 91 per cent (51 of 56 total patients). The resultant function, applied prospectively to 17 additional cases, classified all of them correctly. In patients with cholelithiasis and gallbladders visualized on oral cholecystography, discriminant analysis can improve the prediction of gallstone composition and the subsequent selection of medial or surgical therapy.

Adult↗
PubMedDate not supplied

Discriminant analysis of biochemical parameters in liver disease.

Discriminant function analysis has been used to investigate the relative value of six biochemical parameters (plasma ferritin, C-reactive-protein, bilirubin, alkaline phosphatase, glutamic oxaloacetic acid transaminase and albumin) in the diagnosis of liver disease. This was done among four groups totalling 70 subjects including healthy controls and patients with acute viral hepatitis, liver cirrhosis and primary hepatocellular carcinoma. Albumin had most value in distinguishing between groups, followed cumulatively by ferritin, alkaline phosphatase, C-reactive protein, bilirubin and glutamic oxaloacetic acid transaminase. However, if data on albumin, alkaline phosphatase, bilirubin and glutamic oxaloacetic acid transaminase had already been routinely collected, there would be no advantage in collecting data on ferritin and C-reactive protein. Any four of the six parameters would be of about equal value in distinguishing between diagnostic groups. When the data on all six biochemical parameters was combined in an optimum way, about 66% of all individuals could be correctly assigned to one of the four groups using biochemical markers alone. While the control subjects and patients with acute viral hepatitis formed a relatively well defined, tight cluster (apart from two patients with acute viral hepatitis), patients with liver cirrhosis and primary hepatocellular carcinoma were almost indistinguishable, using these biochemical parameters. If the latter two groups were pooled, then about 86% of subjects could be correctly classified.

Alkaline Phosphatase↗
PubMedDate not supplied

Artificial neural networks and linear discriminant analysis: a valuable combination in the selection of new antibacterial compounds.

A set of topological descriptors has been used to discriminate between antibacterial and nonantibacterial drugs. Topological descriptors are simple integers calculated from the molecular structure represented in SMILES format. The methods used for antibacterial activity discrimination were linear discriminant analysis (LDA) and artificial neural networks of a multilayer perceptron (MLP) type. The following plot frequency distribution diagrams were used: a function of the number of drugs within a value interval of the discriminant function and the output value of the neural network versus these values. Pharmacological distribution diagrams (PDD) were used as a visualizing technique for the identification of antibacterial agents. The results confirmed the discriminative capacity of the topological descriptors proposed. The combined use of LDA and MLP in the guided search and the selection of new structures with theoretical antibacterial activity proved highly effective, as shown by the in vitro activity and toxicity assays conducted.

Anti-Bacterial Agents↗
PubMed1989

Discriminant analysis of trace element distribution in normal and malignant human tissues.

Discriminant analysis of 16 trace element levels measured by ultramicro energy dispersive X-ray fluorescence in malignant and histologically normal human breast, colon, and lung tissues is shown to be a potentially valuable methodology for making malignant-normal and tissue-type classifications. Linear composites of trace elements producing optimal malignant-normal discriminations are found to differ with respect to the number and identity of elements included in the composite for breast, colon, and lung tissues. Nine-, 10-, and 11-element discriminant functions produced overall classification accuracies of 98% for breast, 100% for colon, and 100% for lung tissues, respectively. Elements found to be most important in distinguishing between malignant and normal tissues are Ca, Rb, and Zn in breast, Ca, Zn, and Fe in colon, and Fe, Mn, and Cu in lung samples. Three-group discriminations between breast, colon, and lung tissues were 85% accurate using trace element levels in paired malignant-normal tissues and 91% accurate using trace element levels in tumor tissues only.

Humans↗
PubMed1982

Discriminant analysis of volume-time and flow-volume parameters between healthy adults and asthmatic patients.

Volume-time (V-T) and flow-volume (F-V) curves were measured in all the subjects of nonsmoking young males (mean value 26.3 yrs.), healthy and asthmatic. Eleven parameters of pulmonary function tests, which were composed of two V-T, six F-V, and three mean time constants (MTC) parameters, were calculated from the curves. These were used in the discriminant analysis through all possible selection procedures (APSP) to make clear the importance of the F-V recognition. In using only one parameter, V75, which was one of the F-V parameters, showed the lowest probability of misclassification, 18.78%, and was the most useful parameter to discriminate the two groups. The probability of misclassification of the eleven parameters showed 15.46%, and that of the six F-V ones showed 17.45%. Though the probability of the six F-V ones was higher than that of the eleven ones, it was lower than that of the twoV-T or of the three MTC ones. Therefore the six F-V parameters including V75 was sufficient to discriminate the two groups of subjects. Thus it was made clear that the flow-volume recognition was important by the discriminant analysis.

Adult↗
PubMed1992

Discriminant analysis of the standard 12-lead ECG for diagnosing non-Q wave myocardial infarction.

Discriminant analysis was performed on 12 standard lead data from 159 normal subjects (N) and 304 patients with first myocardial infarction (MI): the latter group consisted of 543 patients with acute non-Q wave MI (NQMI-group A), 68 patients with acute Q wave MI (QMI-group B) and 183 patients (group C) with recent (29) or old (154) QMI. A discriminant function was computed to separate optimally the larger group of QMI patients (group C) from N. A total of 7 features accounted for a specificity of 92% and a sensitivity of 89%. The classification model was then tested on patients with acute MI, regardless of the presence of Q waves (groups A and B); rates of correct classification were 72% for acute NQMI and 85% for acute QMI. The best measurements were voltages in the late portion of the T wave in aVR, V1 and V5, in early and late QRS in V2, at mid-QRS in lead II and in the second half of the P wave in V1. A weighted combination of these features with the coefficients of the discriminant function produced individual discriminant scores for each subject. Group-mean scores were 1.82 for N, -1.27 for acute QMI, -1.14 for old QMI and -.44 for acute NQMI, indicating that acute NQMI was "closer" to N than both acute and old QMI. QRS measurements from the 12-lead ECG were also used to derive the 45 criteria/33 point Selvester score in 53 patients with NQMI: 32% of NQMI were classified as MI with a score of 3 points or more (corresponding to a posterior probability greater than .50). These results were compared with those achieved by multivariate analysis using only QRS measurements: 56% of NQMI were classified as MI with a posterior probability threshold greater than .50. Associating a point score greater than or equal to 1 with criteria for ST-T abnormalities yielded a sensitivity of 72% at a specificity level of 95%. The results emphasize the presence of diagnostic information outside the initial part of QRS, the power of multivariate statistical procedures applied on continuous measurements and the potential benefit of discriminant scores for quantitative assessment of myocardial infarction.

Discriminant Analysis↗
PubMed1989

[Quantitative evaluation of the exercise test using discriminant analysis].

One hundred and twenty seven selected patients had treadmill tests and coronary angiograms. Eight variables were correlated by discriminant analysis to the presence and extent of coronary artery disease (CAD) in a two-way approach. In the first (43 patients without significant CAD versus 84 with significant CAD), two ergometric (maximal heart rate achieved and exercise time) and two common (age and sex) variables merged as predictors. In the second approach (65 patients without CAD or CAD with a little amount of myocardium at risk versus 62 with CAD and a greater extent of jeopardized myocardium), the two clinical variables were replaced by two new ergometric predictors: the level of ST depression and maximum systolic blood pressure reached. Comparatively with the conventional assay of exercise test, the second discriminant function (Z2) enhanced specificity (67 versus 84%) but impaired sensibility (82 versus 70%); nevertheless, this late index was preserved (93 versus 89%) in the 3-vessel CAD group. Moreover, the Z2 value decreased as the severity of CAD increased; the differences between groups were significant (p less than 0.01) up to the level of 2-vessel versus 3-vessel CAD where the trend persists but failed to be significant. Likewise, within the 1-vessel CAD group, the mean Z2 value was lower (p less than 0.01) when the narrowing was located in the proximal portion of the left anterior descending coronary artery. This quantitative approach to the exercise test was validated in a new series of 85 patients and, in our opinion, could be useful for making patient care decisions.

Angiocardiography↗
PubMed1978

Optimum laboratory test combinations for thyroid function studies, selected by discriminant analysis.

In 430 patients with a variety of thyroid disorders, linear discriminant analysis was used to select laboratory test combinations giving optimum diagnostic efficiency in thyroid function studies. Unexpectedly, TSH was found valuable in the diagnosis of hyperthyroidism and T3 in hypothyroidism. These test combinations were found optimal for the separation of euthyroidism/hyperthyroidism: T3, TSH and T4; euthyroidism/hypothyroidism: Combined free thyroid hormone index (FTI) and TSH; and hyperthyroidism/euthyroidism/hypothyroidism: T3, TSH, T4, FTI and T3U. The latter test combination had a total efficiency of 94%. Cholesterol, achilles reflex time, PBI and radioiodine uptake measurement contributed little to the discrimination.

Adolescent↗
PubMed1985

Discriminant analysis of thallium-201 myocardial scintigrams.

In a group of 207 patients undergoing coronary angiography, myocardial scintigrams have been performed at rest (n = 95), after exercise testing (n = 50) or after dipyridamole vasodilation (n = 62). Pictures recorded in anterior and 45 degrees left anterior oblique projections have been analysed according to a model which divides each projection into 5 territories. Thallium uptake has been visually assessed using a three grade scale: 0 = normal uptake, + = sightly abnormal uptake, ++ = evident hypoactivity. With reference to angiography, sensitivity and specificity for prediction of coronary artery disease have been calculated in a classical way using three different criteria of positivity of increasing severity, and by a computerized method based on discriminant analysis. In the first case, sensitivity and specificity were highly dependent one on the other and, as could be expected, varied in opposite directions, a high sensitivity (89 to 100%) being only achieved by accepting a lack of specificity (33 to 57%). Discriminant analysis, on the contrary, provided for each type of protocol: rest, stress, dipyridamole and redistribution, a single optimized combination of sensitivity (74 to 87%) and specificity (92 to 100%). The corresponding point, when located on the receiver operating characteristics (ROC) diagram, demonstrated a clear improvement in diagnostic accuracy. The same method has been applied for discrimination, not only between normals and abnormals, but also between normals and patients with and without myocardial infarction (correct diagnosis in 50 to 68% of cases), and between normals and patients with single-, double- or triple-vessel disease (correct diagnosis in 17 to 65% of cases) with a high specificity (correct classification of normals in 86 to 100% of cases).

Coronary Angiography↗
PubMed1988

Coronary atherosclerotic lesion: its characterization applying an atherometric system, using discriminant analysis.

A given set of autopsies (total 2043) divided into high and low atherosclerosis groups (HAG, LAG) according to the primary cause of death 1171 and 872 cases, respectively, was studied by an atherometric system. The variables of this system characterize the pathomorphological changes, i.e. fatty streaks (X), fibrous plaques (Y) and severe (complicated and calcified) plaques (Z) of atherosclerosis. In the present study were also included two indices, stenosis (P) and benignity (B), applicable for estimation of the severity of the process. In order to verify the discrimination between the HAG and LAG groups the utility of the classification should be demonstrated. The three main branches of the coronary arteries, right (RC), left anterior descending (LAD) and left circumflex (LC), were examined. Qualitative and quantitative analyses of the different vascular changes were made by means of a digitizer coupled to a personal computer NEC 9801 (Japan). The data were processed in a computer EC-1040 (GDR). The multivariate statistical techniques, including discriminant analysis were used applying the "SPSS" commercial statistical package programme. The atherometric system proved to be useful to distinguish between HAG and LAG. Correct classification was in all cases greater than 70% except for LC (68%). The three variables X, Y and Z were capable to separate the groups. The stenosis and benignity indices proved to be the most effective for discrimination. The sign of benignity index discriminating function is always the same of the LAG. This variable is indeed one to indicate benignity. The coherence and consistency of the atherometric system developed was proved by discriminant analysis.

Coronary Artery Disease↗
PubMed1980

Discriminant analysis of ribosomal protein synthesis findings in carrier detection of Duchenne muscular dystrophy.

Previous studies have shown that in vitro muscle ribosomal protein synthesis (RPS) by monomeric ribosomes (MR) and total polyribosomes (TPR) and collagen synthesis (CS) are significantly increased (P less than 0.01) in 47 known carriers of Duchenne muscular dystrophy as compared to 60 age-matched controls. However, there was considerable overlap of the distribution of controls and carriers, particularly for monomeric ribosomal protein synthesis and collagen synthesis. To improve detection of carriers we used discriminant analysis utilizing logs of each measurement as superior to a univariate or bivariate scheme. This study considered four groups: proven carriers (30) (group 1), presumptive carriers (female relatives of Duchenne patients with high serum creatine kinase (CK) levels) (32) (group 2), controls greater than or equal to 20 years old (42) (group 3) and controls less than 20 years (36) (group 4). Comparison of groups, Misclassification (%), (see chart). These results suggest that discriminant analysis reduces the misclassification rates as compared with univariate or bivariate analysis and confirm the superiority of RPS measurements as a carrier test for Duchenne muscular dystrophy.

Adult↗