Some misuses of discriminant analysis.
Explore the source record for details and available documents.
Adult↗
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.
[Discriminative analysis in detecting cardiac insufficiency in myocardial infarct on the basis of the values of stroke volume and blood lactic acid level].
In 203 patients with large focal and transmural myocardial infarctions the cardiac output was studied by means of the dye dilution method, as well as the content of the lactic (Barker and Summerson's technique) and pyruvic acids (Travina's technique). Along with the growth of cardiac insufficiency a fall was observed in the cardiac and stroke indices, and a growing content of the lactic and pyruvic acids and of the L/P factor. The stroke index and the blood lactate level proved to be the most informative values. With a stroke index below 24 ml/m2 all patients with myocardial infarction displayed signs of cardiac insufficiency, while with a stroke index above 54.2 ml/m2 cardiac insufficiency was lacking. A stroke index falling between 24 and 54.2 ml/m2 was observed in both the patients with cardiac insufficiency, and in those without it. A lactic acid content above 31.8 mg% was always concomitant with cardiac insufficiency. With the lactic acid level ranging from 16 to 31.8 mg% the probability of cardiac insufficiency was doubtful.
Acute Disease↗
Data mining techniques applied to medical information.
Knowledge discovery from the dramatically increased data of an auto-stored medical information system is still in its infancy. The purpose of this study is to use widely available and easily operated techniques that can satisfy general users in extracting specific knowledge to make the medical information system more functional. Data mining techniques, including data visualisation, correlation analysis, discriminant analysis, and neural networks supervised classification, were applied to heart disease databases. These techniques can help to identify high risk patients, define the most important factors (variables) in heart disease, and build a multivariate relationship model to show the relationship between any two variables in a way that such relationships are easy to view. Simple visualization techniques were utilised to construct this model, which corresponds with current medical knowledge. Two nonparametric (distribution assumption free) classification tools were employed to identify high risk heart disease patients. Both the neural networks supervised classification methods and the discriminant analysis method produced reliable classification rates for heart disease patients. However, neural networks yielded a higher percentage of correct classifications (averaging 89%) than discriminant analysis (79%). Data visualisation and correlation analysis resulted in similar conclusions regarding the most important factors in heart disease. These data mining tools provide simple and effective methods of extracting knowledge from general medical information. The treatment of missing data is also discussed.
Computer Graphics↗
Determination of volatile compounds in cows' milk using headspace GC-MS.
The composition of the volatile fraction of milk from cows was investigated in a survey of milk samples using a headspace sampling technique and gas chromatography coupled to mass spectrometry analysis (GC-MS). Milk samples were collected from 12 farms, selected for similar management, breed and level of production. Farms were also grouped according to the type of forage in the ration: (1) hay; (2) hay and maize silage; (3) hay, maize silage and grass silages. Forty-one compounds in milk were isolated and identified from GC-MS headspace analysis. Quantitatively, the most representative chemical class was ketones (eight compounds, 170 microg/kg), followed by aldehydes (nine compounds, 63 microg/kg), alcohols (eight compounds, 36 microg/kg), and lower amounts of hydrocarbons (six compounds), sulphur compounds (three compounds), esters (four compounds) and terpenes (three compounds). The novel headspace sampling technique, and the consequent reduction of sample pre-treatment, allowed the identification of low-molecular weight volatile compounds, and reduced the risk of producing artefacts during analysis. Discriminant analysis was used to identify a classification criterion for milk samples, using type of forage in the ration as a grouping variable. Posterior probability error rate indicated that aldehydes provided one of the best discriminant criteria for grouping milks according to ration composition. When all 41 identified volatile compounds were included, discriminant analysis selected nine compounds (acetone, 2,3-butanedione, 2-butanone, ethanol, acetaldehyde, ethylacetate, ethvlisovalerate, dimethylsulphone) that did not fail the tolerance test and which correctly classified 100% of the original cases.
Alcohols↗
Characterization of Spanish unifloral honeys by solid phase microextraction and gas chromatography-mass spectrometry.
Volatile compounds have been investigated in unifloral honeys of the most popular types in Spain. A total of 21 eucalyptus samples, 35 rosemary samples, 33 heather samples, and 15 citrus samples were collected in the course of 3 years in different Spanish regions. Samples were analyzed by SPME followed by GC-MS. About 83 compounds were identified; the concentrations of 33 of them were selected to be processed by multivariate analysis. Discriminant analysis allowed correct assignment of most samples: 94% citrus, 92% eucalyptus, 84% heather, and 84% rosemary. As pure reference honeys are not available, a two step data analysis is proposed, selecting the samples most clearly classified in discriminant analysis as "reference samples" to be used in multiple regression to estimate the most representative compounds for each honey type and considering the rest of the samples as of mixed origin.
Gas Chromatography-Mass Spectrometry↗
Clinical profiles in fibromyalgia patients of the community mental health center: a predictive index of psychopathological severity.
INTRODUCTION: In recent years we have seen an increasing demand for mental health care in patients with fibromyalgia and psychiatric symptoms, although it is not clear if the symptoms are primary or secondary to the presence of the syndrome. This fact has led mental health providers to think that there would be some psychological factors influencing the vulnerability of suffering this painful syndrome, because its etiology is quite non-specific. Bradley et al. (1978) identified different psychopathological profiles within chronic pain syndromes with the MMPI, which were subsequently adapted by Yunus et al. (1991) for fibromyalgia. This present work studied the clinical profile in patients with fibromyalgia. METHOD SAMPLE: 75 patients with fibromyalgia from the community mental health center and 55 healthy subjects. Tools: STAI-E/R, BDI, MMPI-2, MMPI-2 personality disorders, MMPI-2 PSY-5. STATISTICAL ANALYSIS: descriptive statistics and mean comparison (Student's t test). Confirmatory cluster analysis. Discriminative analysis of subgroups. RESULTS: Two different patterns were obtained: group A (32 %) with a typical chronic pain profile (CP) and group B (68 %) with a psychological maladjustment profile (PM). With the discriminative analysis, we obtained the coefficients of the discriminative canonical functions that maximize the differences between both groups. CONCLUSIONS: We confirmed Bradley's classification, obtaining two different psychopathological patterns in the fibromyalgia syndrome sample we studied. We obtained an index of psychopathological profile in fibromyalgia, which would form a new scale, from MMPI-2 for discriminating psychopathological severity in fibromyalgia.
Adult↗
Applying pattern recognition methods and structure property correlations to determine drug carrier potential of nicotinic acid and analogize to dihydropyridine.
Multivariate methods are utilized to compare nicotinic acid and dihydropyridine as a drug carrier. Nicotinic acid and dihydropyridine form ester groups on 10 beta-lactam antibiotics with an oxymethyl group forming a linkage between the antibiotic and the drug carrier (nicotinic acid or dihydropyridine). Calculated molecular properties are analyzed by self-organizing tree algorithm (SOTA), bivariate regression, cluster analysis, factor analysis, discriminant analysis, hierarchical classification, and principal coordinates analysis. Ten important pharmacological properties for each of the nicotinic acid and dihydropyridine antibiotic derivatives are numerically similar and highly correlated. Calculated molecular properties include molar refractivity, molar volume, parachor, index of refraction, partition coefficient (log P), polarizability, and polar surface area. Dermal permeability coefficients (Kp) for nicotinic acid derivatives are similar to values for dihydropyridine derivatives. Dermal permeabliity coefficients analyzed by hierarchical classification and SOTA analysis were shown to be closely interrelated and highly correlated. Ten properties of the nicotinic acid and dihydropyridine were compared by Passing-Bablok regression analysis and shown to be highly correlated (r=0.9879). Box plot analysis of 10 properties, inclusive of both groups of derivatives, showed narrow ranges in values. Cluster analysis of derivative properties showed the nicotinic acid derivatives to be highly similar to the dihydropyridine derivatives of the same antibiotics. Cluster analysis was performed by single linkage, complete linkage, and centroid linkage. Factor analysis showed the nicotinic acid derivatives to be interrelated and similar to the dihydropyridine derivatives. Discriminant analysis performed on all derivatives formed a single highly cohesive and non-differentiated cluster, demonstrating strong similarity among nicotinic acid and dihydropyridine derivatives. Principal coordinates analysis (determines similarity) of Kp values showed high similarity between the nicotinic acid and dihydropyridine antibiotic derivatives.
Dihydropyridines↗
Applying pattern recognition methods to analyze the molecular properties of a homologous series of nitrogen mustard agents.
The purpose of this research was to analyze the pharmacological properties of a homologous series of nitrogen mustard (N-mustard) agents formed after inserting 1 to 9 methylene groups (-CH2-) between 2 -N(CH2CH2Cl)2 groups. These compounds were shown to have significant correlations and associations in their properties after analysis by pattern recognition methods including hierarchical classification, cluster analysis, nonmetric multi-dimensional scaling (MDS), detrended correspondence analysis, K-means cluster analysis, discriminant analysis, and self-organizing tree algorithm (SOTA) analysis. Detrended correspondence analysis showed a linear-like association of the 9 homologs, and hierarchical classification showed that each homolog had great similarity to at least one other member of the series-as did cluster analysis using paired-group distance measure. Nonmetric multi-dimensional scaling was able to discriminate homologs 2 and 3 (by number of methylene groups) from homologs 4, 5, and 6 as a group, and from homologs 7, 8, and 9 as a group. Discriminant analysis, K-means cluster analysis, and hierarchical classification distinguished the high molecular weight homologs from low molecular weight homologs. As the number of methylene groups increased the aqueous solubility decreased, dermal permeation coefficient increased, Log P increased, molar volume increased, parachor increased, and index of refraction decreased. Application of pattern recognition methods discerned useful interrelationships within the homologous series that will determine specific and beneficial clinical applications for each homolog and methods of administration.
Antineoplastic Agents, Alkylating↗
[Discriminant analyses for pregnancy-related changes in hematological and serum biochemical values in cynomolgus monkeys (Macaca fascicularis)].
Hematological and serum biochemical data obtained from 45 non-pregnant and 142 pregnant cynomolgus monkeys (Macaca fascicularis) were analyzed by discriminant analyses. All animals used had been reared under uniform environmental conditions at Tsukuba Primate Center for Medical Science, N. I. H., Japan. The examination items were as follows: red blood cell count(RBC), hematocrit value(Ht), hemoglobin concentration(Hb), mean corpuscular volume(MCV), white blood cell count(WBC), total protein (TP), albumin (ALB), glucose (GLU), total cholesterol (TCHO), free cholesterol (FCHO), triglyceride (TG), and albumin-globulin ratio (A/G). The pregnant animals were divided into four groups according to their gestation ages (Group I: 1-40 days; II: 41-80 days; III: 81-120 days; and IV: more than 121 days). During pregnancy period (Groups II, III and IV), significant decrease of TP, ALB, TCHO, FCHO and TG was observed. RBC, Ht and Hb decreased in the last of the four periods of pregnancy (Group IV). Discrimination was possible between each of these four pregnant groups and the non-pregnant group on the basis of the Mahalanobis' generalized distance. To clarify the relationship among groups, canonical discriminant analysis (discriminant analysis with reduction of dimensionality) was carried out. As a result, the non-pregnant group and the pregnant I group could be clearly discriminated from the other pregnant groups (II, III, and IV) by the first canonical variate. Judging from the relative eigenvector value of each variable for the first canonical variate, the effective discriminant variables were RBC, Ht, Hb, MCV, TP and A/G.(ABSTRACT TRUNCATED AT 250 WORDS)
Analysis of Variance↗
A combination of cytological and architectural morphometry in assessing regenerative hyperplasia and dysplasia in ulcerative colitis.
Morphometry and stereology were used to assess the cytological and architectural characteristics of regenerative and dysplastic colorectal mucosal abnormalities in ulcerative colitis. Reproducibility of measurements was assessed and found to be good, confirming the objective and reproducible nature of morphometric analysis. Discriminant analysis between the morphometric features of regenerative mucosal change and low-grade dysplasia revealed the significant cytological and architectural variables from which a cytological and architectural score were derived. The architectural score was capable of classifying all of the cases into their correct diagnostic category, although a combination of the two scores provided better separation of cases. Probability density analysis was carried out so that probabilities of group membership could be allocated to cases, given their cytological and architectural scores. Discriminant analysis was also applied to low- and high-grade dysplasia. Important cytological and architectural variables were identified and used separately to derive scoring systems. In combination, the dual scoring of lesions was capable of separating low- from high-grade dysplasia, although overlap was still evident. Again, probability density analysis allowed the allocation of cases into one or other category, although a closer examination showed that such a rule could not successfully classify a new set of low- and high-grade dysplasia cases. Quantitative histological analysis of mucosal abnormalities is shown to be of use in the objective diagnosis of reactive and dysplastic change in patients with ulcerative colitis. The use of simple scoring systems and probability based allocation of cases promotes the future role of morphometric techniques in the diagnostic laboratory.
Bayes Theorem↗
The role of vision on hand preshaping during reach to grasp.
During reaching to grasp objects with different shapes hand posture is molded gradually to the object's contours. The present study examined the extent to which the temporal evolution of hand posture depends on continuous visual feedback. We asked subjects to reach and grasp objects with different shapes under five vision conditions (VCs). Subjects wore liquid crystal spectacles that occluded vision at four different latencies from onset of the reach. As a control, full-vision trials (VC5) were interspersed among the blocked vision trials. Object shapes and all VCs were presented to the subjects in random order. Hand posture was measured by 15 sensors embedded in a glove. Linear regression analysis, discriminant analysis, and information theory were used to assess the effect of removing vision on the temporal evolution of hand shape. We found that reach duration increased when vision was occluded early in the reach. This was caused primarily by a slower approach of the hand toward the object near the end of the reach. However, vision condition did not have a significant effect on the covariation patterns of joint rotations, indicating that the gradual evolution of hand posture occurs in a similar fashion regardless of vision. Discriminant analysis further supported this interpretation, as the extent to which hand posture resembled object shape and the rate at which hand posture discrimination occurred throughout the movement were similar across vision conditions. These results extend previous observations on memory-guided reaches by showing that continuous visual feedback of the hand and/or object is not necessary to allow the hand to gradually conform to object contours.
Adult↗
Evaluation of pyridine-3-carboxylic acid as a drug carrier by utilizing multivariate methods, structure property correlations, and pattern recognition techniques.
Multivariate methods and molecular properties are utilized to show similarity of pyridine-3-carboxylic acid (nicotinic acid) to seven drugs that penetrate the central nervous system. Multivariate methods applied include cluster analysis, discriminant analysis, correspondence analysis, self organizing tree algorithm (SOTA) analysis, factor analysis, and principal component analysis. Numerical values of properties for nicotinic acid showed very high correlation with the values from dihydropyridine, barbital, metharbital, phenobarbital, methohexital, 4-aminohex-5-enoic acid, and (4-chlorophenyl)(5-fluoro-2-hydroxyphenyl)methanone. Descriptive statistics of property values for these drugs showed overlapping numerical values for partition coefficients, index of refraction, and nOnN. Principal component analysis and factor analysis of molecular properties showed that nicotinic acid is highly similar to dihydropyridine as well as to other members of this group of drugs. Applying cluster analysis utilizing standard euclidean distance with single linkage and centroid linkage showed that nicotinic acid is highly similar to dihydropyridine and both are consistently grouped into the identical cluster (indicating high similarity). Both nicotinic acid and dihydropyridine show zero violations of the Rule of 5, which indicates good bioavailability characteristics. Discriminant analysis of molecular properties for these eight compounds could not demonstrate differentiation between nicotinic acid and dihydropyridine within the properties applied, this indicating high level of similarity. Neural cluster analysis of molecular properties showed that nicotinic acid and dihydropyridine are included into the same cluster (indicating very strong similarity). SOTA analysis also placed nicotinic acid and dihydropyridine into the same cluster unit. SOTA analysis of molecular properties indicates high similarity between nicotinic acid and dihydropyridine. Correspondence analysis was performed on the molecular properties and showed that there exists considerable association between dihydropyridine and nicotinic acid. Multiple regression analysis of these molecular properties produced a mathematical equation to predict the formula weight of similar compounds for use as drug carriers.
Algorithms↗
Cluster analysis as a method for determining size ranges for spinal implants: disc lumbar replacement prosthesis dimensions from magnetic resonance images.
STUDY DESIGN: Statistical analysis of clinical radiologic data. OBJECTIVE: To develop an objective method for finding the number of sizes for a lumbar disc replacement. SUMMARY OF BACKGROUND DATA: Cluster analysis is a well-established technique for sorting observations into clusters so that the "similarity level" is maximal if they belong to the same cluster and minimal otherwise. METHODS: Magnetic resonance scans from 69 patients, with no abnormal discs, yielded 206 sagittal and transverse images of 206 discs (levels L3-L4-L5-S1). Anteroposterior and lateral dimensions were measured from vertebral margins on transverse images; disc heights were measured from sagittal images. Hierarchical cluster analysis was performed to determine the number of clusters followed by nonhierarchical (K-means) cluster analysis. Discriminant analysis was used to determine how well the clusters could be used to classify an observation. RESULTS: The most successful method of clustering the data involved the following parameters: anteroposterior dimension; lateral dimension (both were the mean of results from the superior and inferior margins of a vertebral body, measured on transverse images); and maximum disc height (from a midsagittal image). These were grouped into 7 clusters so that a discriminant analysis was capable of correctly classifying 97.1% of the observations. The mean and standard deviations for the parameter values in each cluster were determined. CONCLUSIONS: Cluster analysis has been successfully used to find the dimensions of the minimum number of prosthesis sizes required to replace L3-L4 to L5-S1 discs; the range of sizes would enable them to be used at higher lumbar levels in some patients.
Adult↗
Using verbal fluency to detect very mild dementia of the Alzheimer type.
The study sample included 76 healthy older adults and 77 individuals with very mild dementia of the Alzheimer type (DAT). Semantic (animal) and letter (S and P) fluency tasks were used to examine quantitative (word generation) and qualitative (category clustering and switching) aspects of verbal fluency. The goal of the study was to evaluate the utility of fluency tasks in discriminating between healthy aging and very mild DAT. Our results indicated better performance for the healthy group than the DAT group in terms of number of words, number of clusters, number of switches, and size of clusters generated (an exception was clustering on letter S fluency). Clustering and switching variables were significantly correlated with number of words generated and therefore were not included in discriminant analysis. Discriminant analysis revealed that the combination of semantic fluency and narrative recall may be particularly useful in differentiating healthy aging from very mild DAT.
Aged↗
Quantitative characterization of eye tracking dysfunction in schizophrenia.
The purpose of this study was to characterize the nature of the processes that are involved in eye tracking dysfunction (ETD). We identified a combination of quantitative measures that best distinguished qualitatively normal eye tracking from qualitatively abnormal eye tracking, using discriminant analysis. Discriminant scores distinguished schizophrenics with ETD from both schizophrenics with normal eye tracking and normal controls, but did not distinguish schizophrenics with normal eye tracking from normal controls, underscoring the heterogeneity of schizophrenic patients with respect to eye tracking. The results of the discriminant analysis indicated that ETD is a multivariate process involving a primary impairment in the smooth pursuit system characterized by increased catch-up saccades and reduced gain, and, secondarily, disinhibition of intrusive saccades, especially square-wave jerks. Quantitative characterization of ETD makes it possible to consider eye tracking as a quantitative trait in genetic investigations of a multidimensional phenotype.
Adult↗
Fasting serum leptin levels in the analysis of body mass index cut-off values: are they useful for overweight screening in children and adolescents? A school population-based survey in three provinces of central Italy.
OBJECTIVE: Body mass index (BMI) was determined in a population of school students from three provinces of central Italy. Fasting serum leptin concentrations were assayed in a large number of subjects from the same area, to determine their distribution as plotted against the standard deviation score (z-score) of BMI. DESIGN, SUBJECTS AND MEASUREMENTS: Height and weight were recorded from 31170 subjects (16175 male and 14995 female), aged 3-18 y, to construct BMI charts of children and adolescents from central Italy. Percentiles and z-score were calculated using the LMS method of Cole. Serum leptin concentrations were assayed in 1929 subjects (996 male and 933 female) after overnight fasting. RESULTS: BMI percentiles of central Italy were higher than those from standards of other European and USA populations. When plotted against the z-score of BMI, serum leptin values were distributed according to an exponential curve, showing a steep pattern and a wide distribution, as BMI values increased. The hypothesis of the existence of two subgroups, based on a different relation between leptin and BMI, was verified and a separation point between the two subgroups was identified using cluster analysis, discriminant analysis and a novel method developed by our group, hereafter referred to as 'regression clustering'. This method allows identification of the value of the independent variable (z-score of BMI) which can be taken as a separation point. This analysis provided the best results and indicated the following separation points: central Italy standard, z-score = 0.72 (76.4th percentile) for males and z-score = 0.69 (75.5th percentile) for females; French standard (the one suggested for a European population by the European Childhood Obesity Group, ECOG), z-score = 1.46 (92.8th percentile) for males and z-score = 1.96 (97.5th percentile) for females. Similar but variable results were obtained when the same analysis was performed on serum leptin concentration, subdivided according to pubertal development (stage I, stage II-III, stage IV-V). CONCLUSIONS: Children and adolescents from central Italy had greater BMI percentiles when compared to other European populations. Fasting serum leptin concentrations showed a distribution pattern related to z-score, thus allowing to identification of two different subgroups. The z-scores of BMI, identified as separation points, indicated a trend to leptin production by adipocytes that could be taken as indicators of significant increases of fat mass. This study proposes criteria and a statistical approach that could be useful in the identification of BMI cut-off values when screening children and adolescents for overweight.
Adolescent↗
Pathology, diagnosis, and classification of pressure ulcers: comparing clinical and imaging techniques.
Pressure ulcer classification systems are based on the clinical manifestations of the skin and tissue layer affected rather than underlying pathology. The objective of this study was to compare the validity of the clinical grading of erythema (blanching and nonblanching) with a measurement of skin perfusion. Therefore, an exploratory study comparing erythema with laser Doppler imaging of the sacrum and buttock skin areas was undertaken. Acute and major elective general, vascular, and orthopedic surgical in-patients, aged 55 years or over with an expected length of hospital stay of 5 or more days were recruited. Fifty laser Doppler images from 37 patients were obtained and included in a discriminant analysis. Discriminant analysis suggested that blanching and nonblanching erythema were physiologically distinct from "normal" skin; clinically, these could be assessed with reasonable accuracy. Imaging also determined that high blood flow of differing intensity characterized blanching and nonblanching erythema. There was no evidence of the "no flow" phenomenon.
Aged↗
Morphometry and classification of the cat hypothalamic capillaries using multivariate analysis.
Multivariate statistical methods were used to do morphological determination of the amount and the relation of the astrocytic foot processes (AF), astrocytic lamina (AL) and dense zones (DZ) on hypothalamic capillaries. We corroborated that 100% of the perivascular membrane is surrounded by glial astrocytic cytoplasm. Principal component analysis shows that the original variables are independent of one another. Cluster analysis confirms the individual situation obtained through the principal component analysis. Discriminant analysis shows that the discriminant capacity of the 3 variables is very significant. For the potential interrelations of astrocytic foot processes, astrocytic lamina and dense zones we have clearly differentiated four groups of hypothalamic capillaries.
Animals↗