Better Classification (good + classification)

Distribution by Scientific Domains


Selected Abstracts


Protein profile study of breast-tissue homogenates by HPLC-LIF

JOURNAL OF BIOPHOTONICS, Issue 5 2009
K. Kalyan Kumar
Abstract Proteomics is a promising approach for molecular understanding of neoplastic processes including response to treatment. Widely used 2D-gel electrophoresis/Liquid chromatography coupled with mass spectrometry (LC-MS) are time consuming and not cost effective. We have developed a high-sensitivity (femto/subfemtomoles of protein/20 ,l) High Performance Liquid Chromatography-Laser Induced Fluorescence HPLC-LIF instrument for studying protein profiles of biological samples. In this study, we have explored the feasibility of classifying breast tissues by multivariate analysis of chromatographic data. We have analyzed 13 normal, 17 malignant, 5 benign and 4 post-treatment breast-tissue homogenates. Data was analyzed by Principal Component Analysis PCA in both unsupervised and supervised modes on derivative and baseline-corrected chromatograms. Our findings suggest that PCA of derivative chromatograms gives better classification. Thus, the HPLC-LIF instrument is not only suitable for generation of chromatographic data using femto/subfemto moles of proteins but the data can also be used for objective diagnosis via multivariate analysis. Prospectively, identified fractions can be collected and analyzed by biochemical and/or MS methods. (© 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim) [source]


Epothilones: Quantitative Structure Activity Relations Studied by Support Vector Machines and Artificial Neural Networks

MOLECULAR INFORMATICS, Issue 7 2003
Annalen Bleckmann
Abstract In this paper the relation between the structure of epothilones (a new class of anti-tumour agents) and their potential to influence the tubulin-microtubule equilibrium is investigated. Insights into the character of the tubulin-epothilone interactions are derived as the accuracy and reliability of support vector machines and artificial neural networks to model such relations quantitatively is compared. Both methods are well qualified to model relationships between the structure of epothilone derivatives and their anti-tumour activities. Artificial neural networks achieve lower residual standard deviations (22%) compared to support vector machines (25%) and better classification results (89% compared to 75%). However, the reproducibility of the results is greater for support vector machines, which suggests a stronger convergence. The mapping of the influence of individual structural descriptors on the three-dimensional epothilone structure suggests one side of the rather flat molecule to be more important for its activity. The "LIBSVM" software which is used for simulating the support vector machines is freely available from www.csie.ntu.edu.tw/~cjlin/libsvm. The Program "Smart" which is used for simulating artificial neural networks is free for academic use and can be obtained together with the database of epothilones and their activities from www.jens-meiler.de. [source]


Changes in Quality-of-Life After Pacemaker Implantation: Responsiveness of the Aquarel Questionnaire

PACING AND CLINICAL ELECTROPHYSIOLOGY, Issue 3 2001
MONIQUE A. M. STOFMEEL
STOFMEEL, M.A.M., et al.: Changes in Quality-of-Life After Pacemaker Implantation: Responsiveness of the Aquarel Questionnaire. Before being introduced for widespread use, health status instruments should be evaluated for reliability, validity, and responsiveness to relevant clinical changes. In a previous study the validity and reliability of Aquarel, a disease-specific quality-of-life (QOL) questionnaire for pacemaker patients, were tested and found satisfactory. The purpose of this study was to assess the sensitivity to change in health of Aquarel. A cohort of 51 patients was assessed at baseline and at 4,6 weeks after pacemaker implantation. We compared the sensitivity to change over time on the Aquarel scores to the scores on the SF-36 using various techniques (t -test value, effect size, standard error of measurement). Using the 1-standard error of measurement (SEM) criterion for clinically relevant change, Aquarel seemed to provide better classification of patients compared to the SF-36 alone. This study supports the value of Aquarel as a disease-specific measure of QOL in pacemaker patients. [source]


Genetic origins and clinical phenotype of familial and acquired erythrocytosis and thrombocytosis,

AMERICAN JOURNAL OF HEMATOLOGY, Issue 1 2009
Melanie J. Percy
Familial and acquired erythrocytosis and thrombocytosis are characterized by myeloid lineage hyperproliferation, which is either single or multi-lineage in origin. The single lineage disorders exhibit Mendelian inheritance with polyclonal hematopoiesis and often arise from a single genetic defect. In contrast, the multi-lineage disorders exhibit complex patterns of inheritance with multi-genetic origins and clonal hematopoiesis. They have the potential to acquire JAK2 somatic mutations, but this is not the primary event. Identification of the disease-causing genes will enable better classification of familial and acquired erythrocytosis and thrombocytosis. Furthermore, it will provide an insight into the mechanisms regulating myeloid cell proliferation. Am. J. Hematol., 2009. © 2008 Wiley-Liss, Inc. [source]


Neuroimaging advances in holoprosencephaly: Refining the spectrum of the midline malformation,

AMERICAN JOURNAL OF MEDICAL GENETICS, Issue 1 2010
Jin S. Hahn
Abstract Holoprosencephaly (HPE) is a complex congenital brain malformation characterized by failure of the forebrain to bifurcate into two hemispheres, a process normally completed by the fifth week of gestation. Modern high-resolution brain magnetic resonance imaging (MRI) has allowed detailed analysis of the cortical, white matter, and deep gray structural anomalies in HPE in living humans. This has led to better classification of types of HPE, identification of newer subtypes, and understanding of the pathogenesis. Currently, there are four generally accepted subtypes of HPE: alobar, semilobar, lobar, and middle interhemispheric variant. These subtypes are defined primarily by the degree and region of neocortical nonseparation. Rather than there being four discrete subtypes of HPE, we believe that there is a continuum of midline neocortical nonseparation resulting in a spectrum disorder. Many patients with HPE fall within the border zone between the neighboring subtypes. In addition, there are patients with very mild HPE, where the nonseparation is restricted to the preoptic (suprachiasmic) area. In addition to the neocortex, other midline structures such as the thalami, hypothalamic nuclei, and basal ganglia are often nonseparated in HPE. The cortical and subcortical involvements in HPE are thought to occur due to a disruption in the ventral patterning process during development. The severity of the abnormalities in these structures determines the severity of the neurodevelopmental outcome and associated sequelae. © 2010 Wiley-Liss, Inc. [source]


Towards a better classification of erythrokeratodermias

BRITISH JOURNAL OF DERMATOLOGY, Issue 6 2000
D. Hohl
First page of article [source]


Spectral domain OCT of exudative AMD

ACTA OPHTHALMOLOGICA, Issue 2009
N LEVEZIEL
Age-related Macular Degeneration (AMD) is the main cause of vision loss in developed countries. Spectral domain OCT (SD-OCT) is a non invasive technique providing in vivo imaging of the retina, with a higher resolution than time domain OCT. This SIS will describe the clinical features of exudative AMD with SD-OCT, including occult choroidal neovascularization (CNV), classic CNV, idiopathic polypoïdal vasculopathy, and chorioretinal anastomosis. The improvement of the resolution of retinal imaging will provide a better classification and explanation of the pathological processes observed during AMD. [source]


Drug allergy claims in children: from self-reporting to confirmed diagnosis

CLINICAL & EXPERIMENTAL ALLERGY, Issue 1 2008
E. Rebelo Gomes
Summary Background Poorly documented self-reported drug allergy (DAll) is a frequent problem in daily clinical practice and has a considerable impact on prescription choices. The diagnostic work-up of drug hypersensitivity (DHs) allows a better classification of the reactions and provides patients with more reliable information and recommendations for future treatments. Objective To assess the prevalence of self-reported adverse drug reactions (ADRs) and DAll in a paediatric population and to investigate children reporting suspected DAll in order to achieve a firm diagnosis. Design The first phase was based on a cross-sectional survey assessing the life occurrence of ADRs and self-reported DAll carried out at the outpatient clinic of a paediatric hospital. The second phase was based on the diagnostic work-up in children with parent-reported DAll, including detailed anamnesis and in vitro and in vivo investigations (skin and provocation tests). Participants One thousand four hundred and twenty-six parents responded to the initial survey. Sixty of the 67 patients with reported DAll were evaluated at the allergy clinic. Results The prevalences of self-reported ADRs and DAll were 10.2% and 6.0%, respectively. Most of the suspected allergic reactions were non-immediate cutaneous events attributable to ,-lactam antibiotics and occurred in very young children. Thirty-nine of the 60 patients consulting for evaluation had a plausible clinical history and were recommended further investigation. DHs was diagnosed in three children only, based on positive responses in skin (n=1) and oral provocation (n=2) tests. Conclusion ADRs are frequently reported in children, and many children are classified as having a DAll. After complete evaluation, only a few of these reactions can be attributed to DHs and DAll. Most of the patients (94% in this study) could actually tolerate the initially suspected drug. [source]


Report of the ILAE Classification Core Group

EPILEPSIA, Issue 9 2006
Jerome Engel Jr Chair
Summary:, A Core Group of the Task Force on Classification and Terminology has evaluated the lists of epileptic seizure types and epilepsy syndromes approved by the General Assembly in Buenos Aires in 2001, and considered possible alternative systems of classification. No new classification has as yet been proposed. Because the 1981 classification of epileptic seizure types, and the 1989 classification of epilepsy syndromes and epilepsies are generally accepted and workable, they will not be discarded unless, and until, clearly better classifications have been devised, although periodic modifications to the current classifications may be suggested. At this time, however, the Core Group has focused on establishing scientifically rigorous criteria for identification of specific epileptic seizure types and specific epilepsy syndromes as unique diagnostic entities, and is considering an evidence-based approach. The short-term goal is to present a list of seizure types and syndromes to the ILAE Executive Committee for approval as testable working hypotheses, subject to verification, falsification, and revision. This report represents completion of this work. If sufficient evidence subsequently becomes available to disprove any hypothesis, the seizure type or syndrome will be reevaluated and revised or discarded, with Executive Committee approval. The recognition of specific seizure types and syndromes, as well as any change in classification of seizure types and syndromes, therefore, will continue to be an ongoing dynamic process. A major purpose of this approach is to identify research necessary to clarify remaining issues of uncertainty, and to pave the way for new classifications. [source]


Tilting methods for assessing the influence of components in a classifier

JOURNAL OF THE ROYAL STATISTICAL SOCIETY: SERIES B (STATISTICAL METHODOLOGY), Issue 4 2009
Peter Hall
Summary., Many contemporary classifiers are constructed to provide good performance for very high dimensional data. However, an issue that is at least as important as good classification is determining which of the many potential variables provide key information for good decisions. Responding to this issue can help us to determine which aspects of the datagenerating mechanism (e.g. which genes in a genomic study) are of greatest importance in terms of distinguishing between populations. We introduce tilting methods for addressing this problem. We apply weights to the components of data vectors, rather than to the data vectors themselves (as is commonly the case in related work). In addition we tilt in a way that is governed by L2 -distance between weight vectors, rather than by the more commonly used Kullback,Leibler distance. It is shown that this approach, together with the added constraint that the weights should be non-negative, produces an algorithm which eliminates vector components that have little influence on the classification decision. In particular, use of the L2 -distance in this problem produces properties that are reminiscent of those that arise when L1 -penalties are employed to eliminate explanatory variables in very high dimensional prediction problems, e.g. those involving the lasso. We introduce techniques that can be implemented very rapidly, and we show how to use bootstrap methods to assess the accuracy of our variable ranking and variable elimination procedures. [source]


A Bayesian Hierarchical Model for Classification with Selection of Functional Predictors

BIOMETRICS, Issue 2 2010
Hongxiao Zhu
Summary In functional data classification, functional observations are often contaminated by various systematic effects, such as random batch effects caused by device artifacts, or fixed effects caused by sample-related factors. These effects may lead to classification bias and thus should not be neglected. Another issue of concern is the selection of functions when predictors consist of multiple functions, some of which may be redundant. The above issues arise in a real data application where we use fluorescence spectroscopy to detect cervical precancer. In this article, we propose a Bayesian hierarchical model that takes into account random batch effects and selects effective functions among multiple functional predictors. Fixed effects or predictors in nonfunctional form are also included in the model. The dimension of the functional data is reduced through orthonormal basis expansion or functional principal components. For posterior sampling, we use a hybrid Metropolis,Hastings/Gibbs sampler, which suffers slow mixing. An evolutionary Monte Carlo algorithm is applied to improve the mixing. Simulation and real data application show that the proposed model provides accurate selection of functional predictors as well as good classification. [source]


Assessing the color of red wine like a taster's eye

COLOR RESEARCH & APPLICATION, Issue 2 2009
Begoña Hernández
Abstract Color of 33 commercial red wines and five-color reference wines was measured in the same conditions in which visual color assessment is done by wine tasters. Measurements were performed in the two distinctive regions, center and rim, which are the regions assessed by wine tasters when the wine sampler is tilted. Commercial wines were classified into five color categories using the color specifications in their taste cards. The five color categories describe the spread of red hues found in red wines from the violet to brown nuances. The performance of CIELAB color coordinates in terms of their ability to reproduce the observed classification has been established using discriminant analysis. The CIELAB hue angle, hab, measured in the rim, where wine thickness is of the order of few millimeters, gives the best results classifying correctly 71.1% of the samples. Classification results are not significantly improved when additional color coordinates are considered. Moreover, ,E* color differences with color reference wines do not provide good classification results. The analysis of reference and commercial wines supports the fact that hue is the main factor in the classification done by wine tasters. This is reinforced by the linear correlation found between hab in the rim and the wine age (R2 = 0.795) in accordance with the fact that wines change their hues from violet to brown tints with ageing. © 2009 Wiley Periodicals, Inc. Col Res Appl, 34, 153,162, 2009 [source]


Prediction of Tyrosinase Inhibition Activity Using Atom-Based Bilinear Indices

CHEMMEDCHEM, Issue 4 2007
Yovani Marrero-Ponce Prof.
Abstract A set of novel atom-based molecular fingerprints is proposed based on a bilinear map similar to that defined in linear algebra. These molecular descriptors (MDs) are proposed as a new means of molecular parametrization easily calculated from 2D molecular information. The nonstochastic and stochastic molecular indices match molecular structure provided by molecular topology by using the kth nonstochastic and stochastic graph-theoretical electronic-density matrices, Mk and Sk, respectively. Thus, the kth nonstochastic and stochastic bilinear indices are calculated using Mk and Sk as matrix operators of bilinear transformations. Chemical information is coded by using different pair combinations of atomic weightings (mass, polarizability, vdW volume, and electronegativity). The results of QSAR studies of tyrosinase inhibitors using the new MDs and linear discriminant analysis (LDA) demonstrate the ability of the bilinear indices in testing biological properties. A database of 246 structurally diverse tyrosinase inhibitors was assembled. An inactive set of 412 drugs with other clinical uses was used; both active and inactive sets were processed by hierarchical and partitional cluster analyses to design training and predicting sets. Twelve LDA-based QSAR models were obtained, the first six using the nonstochastic total and local bilinear indices and the last six with the stochastic MDs. The discriminant models were applied; globally good classifications of 99.58 and 89.96,% were observed for the best nonstochastic and stochastic bilinear indices models in the training set along with high Matthews correlation coefficients (C) of 0.99 and 0.79, respectively, in the learning set. External prediction sets used to validate the models obtained were correctly classified, with accuracies of 100 and 87.78,%, respectively, yielding C values of 1.00 and 0.73. This subset contains 180 active and inactive compounds not considered to fit the models. A simulated virtual screen demonstrated this approach in searching tyrosinase inhibitors from compounds never considered in either training or predicting series. These fitted models permitted the selection of new cycloartane compounds isolated from herbal plants as new tyrosinase inhibitors. A good correspondence between theoretical and experimental inhibitory effects on tyrosinase was observed; compound CA6 (IC50=1.32,,M) showed higher activity than the reference compounds kojic acid (IC50=16.67,,M) and L -mimosine (IC50=3.68,,M). [source]