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 487 records · Page 27Linked to original sources
PubMed1982

[Discriminant analysis of graphic elements of the EEG. Application to the detection of episodes of diffuse spike-waves].

A database of EEG information was collected from EEG recordings performed in epileptic patients with diffuse spike-wave complex discharges. Normal activity, spike-waves, slow waves and artefacts were mixed up in these recordings. The analysis of EEGs stored in the database was performed, channel by channel, through a 2.56 s moving window. For each so defined EEG sequence, a set of 22 variables chosen for their discriminatory power was computed. A subset of 8 highly discriminating variables was selected by the means of a stepwise discriminant analysis. Each class of the learning set contained 40 up to 100 EEG sequences. A classifying algorithm that takes into account zones of uncertainty is proposed. It has been evaluated on a test set which was composed of 1981 EEG sequences issued from 15 different patients. The results have been checked by two neurologists. The agreement rate between each of them and the proposed algorithm was more than 92%; this result is comparable to the agreement rate between the two neurologists (94%). A contextual analysis algorithm, using bi-dimensional smoothing techniques, allowed to improve the agreement rates which exceeded 94%.

Computers↗
PubMed1991

Neural networks as a tool for utilizing laboratory information: comparison with linear discriminant analysis and with classification and regression trees.

Successful applications of neural network architecture have been described in various fields of science and technology. We have applied one such technique, error back-propagation, to a medical classification problem stemming from clinical chemistry, and we have compared the performance of two different neural networks with results obtained by conventional linear discriminant analysis or by the technique of classification and regression trees. The results obtained by the various models were tested for robustness by jackknife validation ("leave n out" method). Compared with the two other techniques, neural networks show a unique ability to detect features hidden in the input data which are not explicitly formulated as input. Thus, neural network techniques appear promising in the field of clinical chemistry, and their application, particularly in situations with complex data structures, should be investigated with more emphasis.

Alanine Transaminase↗
PubMed2000

Classification of reflectance spectra from pigmented skin lesions, a comparison of multivariate discriminant analysis and artificial neural networks.

Successful treatment of skin cancer, especially melanoma, depends on early detection, but diagnostic accuracy, even by experts, can be as low as 56% so there is an urgent need for a simple, accurate, non-invasive diagnostic tool. In this paper we have compared the performance of an artificial neural network (ANN) and multivariate discriminant analysis (MDA) for the classification of optical reflectance spectra (320 to 1100 nm) from malignant melanoma and benign naevi. The ANN was significantly better than MDA, especially when a larger data set was used, where the classification accuracy was 86.7% for ANN and 72.0% for MDA (p < 0.001). ANN was better at learning new cases than MDA for this particular classification task. This study has confirmed that the convenience of ANNs could lead to the medical community and patients benefiting from the improved diagnostic performance which can be achieved by objective measurement of pigmented skin lesions using spectrophotometry.

Algorithms↗
PubMed1988

Provocation dose and discriminant analysis in histamine bronchoprovocation. Are the current predictive data satisfactory?

In 20 normal subjects (N) and 20 asthmatic patients (A) using bronchodilators as needed, the PD40, PD10, PD15, PD20, PD10T, and PD20C were measured (PD being provocation dose, subscript being -% delta in Gaw/VL for PD40, lowest FEV1 for PD20C, and best FEV1 for the remaining PDs; 10T means delta FEV1 greater than -10 percent). For discriminant analysis we used an ad hoc graphic (best case) method, a ceiling method based on highest PD in A and two methods (logistic and linear) which considered PDs in both N and A (PDN and PDA, respectively). The distribution of PDN and PDA had substantial overlap and appeared log normal. The PD15, PD20, and PD20C displayed the smallest mean misclassification error followed by PD10T, PD40, and PD10. The linear and logistic methods produced balanced sensitivity and specificity but, predictably, a misclassification error higher than that of the graphic method. The ceiling method proved unsatisfactory with 100 percent sensitivity but approximately equal to 60 percent specificity. Using linear and logistic methods, the posttest likelihood of asthma could be expressed as a function of its pretest likelihood and level of PD recorded. We concluded that: (1) the most discriminant PDs are PD20, PD15 and PD20c; and (2) new normative data for diagnostic bronchoprovocation are needed, because: (a) when PDA and PDN overlap, the currently used ceiling method leads to a high misclassification rate, while the linear and logistic method based on mathematical model have a better discriminant ability; (b) to separate PDA from PDN and allow the application of the ceiling method, "as-needed bronchodilators" is not a reproducible criterion.

Adolescent↗
PubMed1992

Distinguishing between conduct-problem and emotionally disturbed students in elementary school: a five-instrument discriminant analysis.

Five instruments, the Children's Manifest Anxiety Scale, Differential Test of Conduct and Emotional Problems, Personality Inventory for Children, Piers-Harris Children's Self-concept Scale, and Revised Behavior Problem Checklist, were employed to effect differentiations between 44 conduct-problem and 31 emotionally disturbed students placed in public elementary school classes for the severely emotionally handicapped. Discriminant analysis of each test's rate of classification germane to both a preassessment categorization as conduct-problem vs emotionally disturbed and a postassessment classification as no problem vs conduct-problem vs emotionally disturbed vs combined problem is presented. The analysis indicated that the Differential Test of Conduct and Emotional Problems, Personality Inventory for Children, and Revised Behavior Problem Checklist can, with variable effectiveness, significantly classify groups using such pre- and postassessment information.

Adolescent↗
PubMed2002

Nonlinear fisher discriminant analysis using a minimum squared error cost function and the orthogonal least squares algorithm.

The nonlinear discriminant function obtained using a minimum squared error cost function can be shown to be directly related to the nonlinear Fisher discriminant (NFD). With the squared error cost function, the orthogonal least squares (OLS) algorithm can be used to find a parsimonious description of the nonlinear discriminant function. Two simple classification techniques will be introduced and tested on a number of real and artificial data sets. The results show that the new classification technique can often perform favourably compared with other state of the art classification techniques.

Algorithms↗
PubMedDate not supplied

Computer discriminant analysis of atypical urothelial cells.

Prior computer studies of digitized cell images by the TICAS system have shown that the category of urothelial cells classified visually as atypical may be composed of 2 subgroups, one clustering mainly with benign cells and the other with malignant cells. As a consequence, a visual review of the group of atypical cells was conducted and tested by computer discriminant analysis. The computer classification confirmed the visual reclassification and subdivision of atypical urothelial cells into 2 subgroups, ATY I and ATY II. This is yet another example of feedback from computer diagnosis to visual assessment of cells. The significance of these observations in terms of diagnosis will be the subject of subsequent communications.

Computers↗
PubMed1988

Delirium in liver transplantation candidates: discriminant analysis of multiple test variables.

We report the results of electroencephalograms, Mini-Mental State exam, Trailmaking Tests A and B, and serum albumin levels in 108 consecutive liver transplantation candidates. We compared test results to a clinical DSM-III diagnosis of delirium. Although each variable could differentiate between the two groups (delirium n = 18; nondelirium n = 90) at a statistically significant level, a discriminant analysis involving either all variables or only three particular variables (Trailmaking B, EEG code, and albumin) resulted in the highest specificity (97.8%) and sensitivity (83.3%), with a correct classification of 95.4% of subjects. The analysis also generates an equation that can be applied to clinical situations to enhance the accurate recognition of delirium. In addition, to explain abnormal Trailmaking B scores and/or EEGs in subjects who did not otherwise meet DSM-III criteria for delirium, we suggest the presence of a "subclinical delirium."

Adolescent↗
PubMed2003

Diagnostic differentiation of essential thrombocythaemia from thrombocythaemias associated with chronic idiopathic myelofibrosis by discriminate analysis of bone marrow features--a clinicopathological study on 272 patients.

Until now diagnosis of essential thrombocythaemia (ET) is generally performed by following the criteria of the Polycythaemia Vera Study Group (PVSG) that only marginally regards morphological features. Bone marrow biopsies were studied from 272 patients with ET in strict accordance with the PVSG guidelines and also from 35 control patients with reactive thrombocytosis. To define morphological features of distinctive impact more accurately, we performed a stepwise discriminant analysis of 16 morphological parameters based on histochemical staining reactions and semiquantitative grading of standardized features. A clear-cut separation into three distinctive histological patterns was accomplished that showed in more than 96% a correct predicted classification. Variables of significant impact included fibre content, quantity and cytological abnormalities of megakaryopoiesis like bulbous (cloud-like) nuclei, degree of nuclear lobulation and presence of giant forms. These changes were not detectable in the control group. The different constellations of histopathological features could be assigned to true ET (98 patients) and false ET, i.e. 136 patients with prefibrotic and 38 patients with early fibrotic chronic idiopathic myelofibrosis (IMF) accompanied by thrombocythaemia. A re-evaluation of clinical findings was in keeping with this classification into three categories that exerted significant differences to develop myelofibrosis during observation time and also different survival patterns. Contrasting IMF true ET is characterized by a pronounced proliferation of the megakaryocyte lineage showing large to giant cells without maturation defects and no relevant increase in reticulin fibres. Discrimination between these entities is warranted, because of a significant difference in presenting haematological data, follow-up and life expectancy.

Bone Marrow Cells↗
PubMed1982

Discriminant analysis of gross and fine motor proficiency data.

Two classes of non-handicapped and two classes of handicapped children (n = 67), aged 5 to 12 years, took part in a study designed to determine whether a regular or specially designed physical education program suited their individual needs best. Initially receiving instruction with class peers each child was administered the 14-item short-form Bruininks-Oseretsky Test of Motor Proficiency. Discriminant analysis of the data indicated that two of the three functions were significant, together accounting for 93% of the variance. For each significant function fewer than half the test items contributed substantial discriminatory power. Over-all, 75% of the subjects were correctly classified but misclassifications occurred in each class; different placements seem appropriate for non-handicapped and handicapped children.

Child↗
PubMedDate not supplied

A multivariate discriminate analysis of behavioral measures in genetically nervous dogs.

For some years we have studied a strain of genetically nervous dogs in the Neuropsychiatric Research Laboratory, Veterans Administration Hospital, North Little Rock, Arkansas. In the manner of Pavlov and Gantt and later Scott and Fuller we have characterized these dogs in such descriptive terms as timid, human aversive, and catatonic-like. Behavioral tests have been administered on nearly all dogs in this longitudinal study, and we are using these data to try to develop statistical procedures to maximize the discriminatory power of the behavioral assay and to more accurately characterize the behavioral deficit. A multivariate discriminate analysis of 13 variables on 91 healthy and 63 nervous dogs assayed at 3 months of age shows: (1) that much of our present behavioral testing procedures is redundant, and (2) that simple "friendliness to humans" in the dog is as effective for discriminating between the two groups as any of the 13 measures, taken either singly or collectively.

Animals↗
PubMed1989

Discriminant analysis of risk factors for sexual victimization among a national sample of college women.

Examined the accuracy with which rape and lesser sexual assaults were predicted among a representative national sample of 2,723 college women. A total of 14 risk variables operationalized three vulnerability hypotheses: (a) vulnerability-creating traumatic experiences, (b) social-psychological vulnerability, and (c) vulnerability-enhancing situations. Each hypothesis was tested individually, and a composite model was developed via discriminant analysis. Only the traumatic experiences variables clearly improved over the base rates in identifying rape victims, but risk variables from each vulnerability hypothesis met criteria for inclusion in the composite model. A risk profile emerged that characterized only 10% of the women, but among them the risk of rape was twice the rate of women without the profile. The concept of traumatic sexualization was used to explain this finding. However, the vast majority of sexually victimized women (75-91%) could not be differentiated from nonvictims.

Adult↗
PubMedDate not supplied

Detection of milk fat adulteration by linear discriminant analysis of fatty acid data.

Analysis of the fatty acid (FA) profile of milk fat (MF) by gas-liquid chromatography is widely used to detect adulteration with foreign fats. On the basis of the FA spectra of 352 genuine Austrian MF samples collected over a 4-year period, the effectiveness of concentration ranges of the major FA of MF and of certain FA ratios to identify non-MF/MF mixtures was tested. FA ratios proved useful for the detection of coconut fat in MF and admixture of vegetable oils rich in linoleic acid down to a level of 2%. This approach failed to identify non-MF/MF blends containing beef tallow, lard, olive oil, or palm oil at a level less than 10% commingling. Linear discriminant analysis applied to FA data was successful in distinguishing pure MF from adulterated MF. Computer-simulated data were used to derive the discriminant functions. Saturated and unsaturated FA with 18 C atoms were the most useful discriminating variables selected by a stepwise variable selection procedure. More than 95% of a data set composed of pure MF, and non-MF/MF blends containing 3% of either tallow, lard, olive oil, or palm oil were correctly classified. The validity of the classification rule was also tested by 206 gravimetrically prepared fat mixtures. Mixtures containing > 3% foreign fat were detected in all cases.

Animals↗
PubMed1983

Discriminant analysis of pulmonary function parameters--mild asthmatics versus moderate asthmatics--.

Maximal expiratory volume-time and flow-volume (MEVT and MEFV) curves were constructed from the measurements of young male nonsmoking, mild and moderate asthmatic patients (mean age, 29.7 yrs.). Eleven parameters of the pulmonary function tests, two MEVT, six MEFV, and three mean time constant (MTC) parameters, were calculated from the curves. These parameters were used in 15 analyses through the all possible selection procedure (APAP) discriminating between mild and moderate asthmatics. The probability of misclassification was computed with each of the eleven parameters, and all eleven probabilities thus obtained were compared with each other. This procedure showed us that the probability of misclassification ranged from 30.83% to 45.40% and that the most useful parameter was MTC50-25. The probability of misclassification computed using all eleven parameters (total parameter group) was 15.90%. The discriminant analysis indicated that the flow-volume patterns varied according the severity of bronchial asthma, thus, the flow-volume curve was considered to be important in analyzing the severity of bronchial asthma.

Adult↗
PubMed1994

Contributions of discriminant analysis to differential diagnosis by self-report.

We used discriminant function analyses of the Minnesota Multiphasic Personality Inventory (MMPI; Hathaway & McKinley, 1983), Millon Clinical Multiaxial Inventory (MCMI; Millon, 1983), MCMI-II (Millon, 1987), and Symptom Checklist Ninety-Revised (SCL-90-R; Derogatis, 1983) profiles from a heterogenous group of 272 psychiatric inpatients to classify patients as depressed, manic, and/or psychotic. Most functions generated from these tests significantly discriminated depressed, manic (not MCMI-II), and psychotic (not MCMI) subjects from psychiatric controls. However, there was little improvement in diagnostic efficiency over the use of single scale elevations at specified cut scores. Functions derived from the MCMI for mania and the MCMI-II for psychosis show the most promise but require replication. The difficulty of using group profile differences for the diagnosis of individual psychiatric patients is discussed.

Adult↗
PubMed2004

Helical CT evaluation of the chemical composition of urinary tract calculi with a discriminant analysis of CT-attenuation values and density.

The aim of this study was to evaluate the efficacy of helical CT using a combination of CT-attenuation values and visual assessment of stone density as well as discriminant linear analysis to predict the chemical composition of urinary calculi. One hundred human urinary calculi were obtained from a stone-analysis laboratory and placed in 20 excised pig kidneys. They were scanned at 80, 120 and 140 kV with 3-mm collimation. Average, highest and lowest CT-attenuation values and CT variability were recorded. The internal calculus structure was assessed using a wide window setting, and visual assessment of stone density was recorded. A stepwise discriminant linear analysis was performed. The following three variables were discriminant: highest CT-attenuation value, visual density, and highest CT-attenuation value/area ratio, all at 80 kV. The probability of correctly classifying stone composition with these three variables was 0.64, ranging from 0.54 for mixed calculi to 0.69 for pure calculi. The probabilities of correctly classifying calculus composition were: 0.91 for calcium oxalate monohydrate and brushite, 0.89 for cystine, 0.85 for uric acid, 0.11 for calcium oxalate dihydrate, 0.10 for hydroxyapatite, and 0.07 for struvite calculi. When the first two ranks of highest probability for the accurate classification of each calculus type were taken into account, 81% of the calculi were correctly classified. Assessment at 80 kV of the highest CT-attenuation value, visual density and the highest CT-attenuation value/area ratio accurately predicts the chemical composition of 64-81% of urinary calculi. When the first two ranks of highest probability for the accurate classification of each calculus type were taken into account, all cystine, calcium oxalate monohydrate and brushite calculi were correctly classified.

Animals↗
PubMed1983

Predictors of the success or failure of medical therapy in patients with chronic recurrent sustained ventricular tachycardia: a discriminant analysis.

To identify predictors of the success or failure of medical therapy in chronic recurrent sustained ventricular tachycardia, univariate and multivariate statistical techniques were used to retrospectively analyze data in 84 patients with this arrhythmia. By univariate analysis, four factors were associated with successful medical treatment: age less than 45 years, ejection fraction greater than 50%, hypokinesia as the only contraction abnormality and the absence of organic heart disease. Four other findings, the induction of ventricular tachycardia with a single ventricular extrastimulus, an HV interval greater than 60 ms, the presence of a left ventricular aneurysm and Q waves on a baseline electrocardiogram, correlated with medical failure. However, none of these variables alone accurately predicted treatment results in more than 75% of cases. By discriminant analysis, a function incorporating eight variables was constructed which correctly classified 81% of patients. Moreover, three-quarters of the patients could be classified into groups with a high or low probability of success where accuracy increased to 90%. When the discriminant function was tested prospectively in 31 similar patients, 25 (81%) fell into the groups with a high or low probability of success. In the latter group, of 20 patients predicted to fail medical therapy, 19 (95%) did fail a complete trial of medical therapy. The overall accuracy remained a high 92%. In clinical application this function would allow patients with a high probability of responding to medical therapy to be selected for serial electrophysiologic drug testing. In patients with a low probability of responding to medical therapy, serial studies could be avoided and alternate forms of therapy explored.

Anti-Arrhythmia Agents↗
PubMed1992

Morphometry and discriminant analysis of the endometrium.

The authors dated 100 normal endometrial biopsies. Only endometrial specimens with histologically confirmed subphases and no evidence of organic disease were accepted for study. The morphometry and stereologic measures of the glands, lumina and endometrial epithelium were assessed. Quantitative assessment of the normal endometrium supported the histologic events that occur in the endometrial cycle (proliferative and secretory phases). There was a progressive increase in morphometric and stereologic values of the glands, epithelium and lumina during the endometrial cycle. Using stepwise discriminant analysis, 94% of the specimens were correctly classified into the categories of proliferative and secretory phase; two quantitative parameters were required, the epithelial volume density and longest luminal diameter. When three subphases within proliferative and secretory endometrium were considered, the overall accuracy was 90% and 78%, respectively.

Biopsy↗