Stratification of the severity of critically ill patients with classification trees

Background: Development of three classification trees (CT) based on the CART (Classification and Regression Trees), CHAID (Chi-Square Automatic Interaction Detection) and C4.5 methodologies for the calculation of probability of hospital mortality; the comparison of the results with the APACHE II, SAPS II and MPM II-24 scores, and with a model based on multiple logistic regression (LR). Methods: Retrospective study of 2864 patients. Random partition (70:30) into a Development Set (DS) n = 1808 and Validation Set (VS) n = 808. Their properties of discrimination are compared with the ROC curve (AUC CI 95%), Percent of correct classification (PCC CI 95%); and the calibration with the Calibration Curve and the Standardized Mortality Ratio (SMR CI 95%). Results: CTs are produced with a different selection of variables and decision rules: CART (5 variables and 8 decision rules), CHAID (7 variables and 15 rules) and C4.5 (6 variables and 10 rules). The common variables were: inotropic therapy, Glasgow, age, (A-a)O2 gradient and antecedent of chronic illness. In VS: all the models achieved acceptable discrimination with AUC above 0.7. CT: CART (0.75(0.71-0.81)), CHAID (0.76(0.72-0.79)) and C4.5 (0.76(0.73-0.80)). PCC: CART (72(69- 75)), CHAID (72(69-75)) and C4.5 (76(73-79)). Calibration (SMR) better in the CT: CART (1.04(0.95-1.31)), CHAID (1.06(0.97-1.15) and C4.5 (1.08(0.98-1.16)). Conclusion: With different methodologies of CTs, trees are generated with different selection of variables and decision rules. The CTs are easy to interpret, and they stratify the risk of hospital mortality. The CTs should be taken into account for the classification of the prognosis of critically ill patients.

Feature Selection and Classification Using Age Layered Population Structure Genetic Programming

The curse of dimensionality is a major problem in the fields of machine learning, data mining and knowledge discovery. Exhaustive search for the most optimal subset of relevant features from a high dimensional dataset is NP hard. Sub–optimal population based stochastic algorithms such as GP and GA are good choices for searching through large search spaces, and are usually more feasible than exhaustive and deterministic search algorithms. On the other hand, population based stochastic algorithms often suffer from premature convergence on mediocre sub–optimal solutions. The Age Layered Population Structure (ALPS) is a novel metaheuristic for overcoming the problem of premature convergence in evolutionary algorithms, and for improving search in the fitness landscape. The ALPS paradigm uses an age–measure to control breeding and competition between individuals in the population. This thesis uses a modification of the ALPS GP strategy called Feature Selection ALPS (FSALPS) for feature subset selection and classification of varied supervised learning tasks. FSALPS uses a novel frequency count system to rank features in the GP population based on evolved feature frequencies. The ranked features are translated into probabilities, which are used to control evolutionary processes such as terminal–symbol selection for the construction of GP trees/sub-trees. The FSALPS metaheuristic continuously refines the feature subset selection process whiles simultaneously evolving efficient classifiers through a non–converging evolutionary process that favors selection of features with high discrimination of class labels. We investigated and compared the performance of canonical GP, ALPS and FSALPS on high–dimensional benchmark classification datasets, including a hyperspectral image. Using Tukey’s HSD ANOVA test at a 95% confidence interval, ALPS and FSALPS dominated canonical GP in evolving smaller but efficient trees with less bloat expressions. FSALPS significantly outperformed canonical GP and ALPS and some reported feature selection strategies in related literature on dimensionality reduction.

Feature Selection and Classification Using Age Layered Population Structure Genetic Programming

The curse of dimensionality is a major problem in the fields of machine learning, data mining and knowledge discovery. Exhaustive search for the most optimal subset of relevant features from a high dimensional dataset is NP hard. Sub–optimal population based stochastic algorithms such as GP and GA are good choices for searching through large search spaces, and are usually more feasible than exhaustive and determinis- tic search algorithms. On the other hand, population based stochastic algorithms often suffer from premature convergence on mediocre sub–optimal solutions. The Age Layered Population Structure (ALPS) is a novel meta–heuristic for overcoming the problem of premature convergence in evolutionary algorithms, and for improving search in the fitness landscape. The ALPS paradigm uses an age–measure to control breeding and competition between individuals in the population. This thesis uses a modification of the ALPS GP strategy called Feature Selection ALPS (FSALPS) for feature subset selection and classification of varied supervised learning tasks. FSALPS uses a novel frequency count system to rank features in the GP population based on evolved feature frequencies. The ranked features are translated into probabilities, which are used to control evolutionary processes such as terminal–symbol selection for the construction of GP trees/sub-trees. The FSALPS meta–heuristic continuously refines the feature subset selection process whiles simultaneously evolving efficient classifiers through a non–converging evolutionary process that favors selection of features with high discrimination of class labels. We investigated and compared the performance of canonical GP, ALPS and FSALPS on high–dimensional benchmark classification datasets, including a hyperspectral image. Using Tukey’s HSD ANOVA test at a 95% confidence interval, ALPS and FSALPS dominated canonical GP in evolving smaller but efficient trees with less bloat expressions. FSALPS significantly outperformed canonical GP and ALPS and some reported feature selection strategies in related literature on dimensionality reduction.

A Characterization of Exact Non-atomic Market Games

Continuous exact non-atomic games are naturally associated to certain operators between Banach spaces. It thus makes sense to study games by means of the corresponding operators. We characterize non-atomic exact market games in terms of the properties of the associated operators. We also prove a separation theorem for weak compact sets of countably additive measures, which is of independent interest.

Non invasive classification system of scoliosis curve types using least-squares support vector machines

Objective To determine scoliosis curve types using non invasive surface acquisition, without prior knowledge from X-ray data. Methods Classification of scoliosis deformities according to curve type is used in the clinical management of scoliotic patients. In this work, we propose a robust system that can determine the scoliosis curve type from non invasive acquisition of the 3D back surface of the patients. The 3D image of the surface of the trunk is divided into patches and local geometric descriptors characterizing the back surface are computed from each patch and constitute the features. We reduce the dimensionality by using principal component analysis and retain 53 components using an overlap criterion combined with the total variance in the observed variables. In this work, a multi-class classifier is built with least-squares support vector machines (LS-SVM). The original LS-SVM formulation was modified by weighting the positive and negative samples differently and a new kernel was designed in order to achieve a robust classifier. The proposed system is validated using data from 165 patients with different scoliosis curve types. The results of our non invasive classification were compared with those obtained by an expert using X-ray images. Results The average rate of successful classification was computed using a leave-one-out cross-validation procedure. The overall accuracy of the system was 95%. As for the correct classification rates per class, we obtained 96%, 84% and 97% for the thoracic, double major and lumbar/thoracolumbar curve types, respectively. Conclusion This study shows that it is possible to find a relationship between the internal deformity and the back surface deformity in scoliosis with machine learning methods. The proposed system uses non invasive surface acquisition, which is safe for the patient as it involves no radiation. Also, the design of a specific kernel improved classification performance.

The classification and application of toxic Chinese Materia Medica

Many important drugs in the Chinese materia medica (CMM) are known to be toxic, and it has long been recognized in classical Chinese medical theory that toxicity can arise directly from the components of a single CMM or may be induced by an interaction between combined CMM. Traditional Chinese Medicine presents a unique set of pharmaceutical theories that include particular methods for processing, combining and decocting, and these techniques contribute to reducing toxicity as well as enhancing efficacy. The current classification of toxic CMM drugs, traditional methods for processing toxic CMM and the prohibited use of certain combinations, is based on traditional experience and ancient texts and monographs, but accumulating evidence increasingly supports their use to eliminate or reduce toxicity. Modern methods are now being used to evaluate the safety of CMM; however, a new system for describing the toxicity of Chinese herbal medicines may need to be established to take into account those herbs whose toxicity is delayed or otherwise hidden, and which have not been incorporated into the traditional classification. This review explains the existing classification and justifies it where appropriate, using experimental results often originally published in Chinese and previously not available outside China.

Lower semicontinuity of attractors for non-autonomous dynamical systems

This paper is concerned with the lower semicontinuity of attractors for semilinear non-autonomous differential equations in Banach spaces. We require the unperturbed attractor to be given as the union of unstable manifolds of time-dependent hyperbolic solutions, generalizing previous results valid only for gradient-like systems in which the hyperbolic solutions are equilibria. The tools employed are a study of the continuity of the local unstable manifolds of the hyperbolic solutions and results on the continuity of the exponential dichotomy of the linearization around each of these solutions.

Strongly damped wave problems: Bootstrapping and regularity of solutions

The aim of the article is to present a unified approach to the existence, uniqueness and regularity of solutions to problems belonging to a class of second order in time semilinear partial differential equations in Banach spaces. Our results are applied next to a number of examples appearing in literature, which fall into the class of strongly damped semilinear wave equations. The present work essentially extends the results on the existence and regularity of solutions to such problems. Previously, these problems have been considered mostly within the Hilbert space setting and with the main part operators being selfadjoint. In this article we present a more general approach, involving sectorial operators in reflexive Banach spaces. (C) 2008 Elsevier Inc. All rights reserved.

Extensions of homogeneous polynomials ON c(0)(l(2)(i))

We show that a 2-homogeneous polynomial on the complex Banach space c(0)(l(2)(i)) is norm attaining if and only if it is finite (i.e, depends only on finite coordinates). As the consequence, we show that there exists a unique norm-preserving extension for norm-attaining 2-homogeneous polynomials on c(0)(l(2)(i)).

UNIFORM APPROXIMATION ON IDEALS OF MULTILINEAR MAPPINGS

For each ideal of multilinear mappings M we explicitly construct a corresponding ideal (a)M such that multilinear forms in (a)M are exactly those which can be approximated, in the uniform norm, by multilinear forms in M. This construction is then applied to finite type, compact, weakly compact and absolutely summing multilinear mappings. It is also proved that the correspondence M bar right arrow (a)M. IS Aron-Berner stability preserving.

Modes of cell death in the hypopharyngeal gland of the honey bee (Apis mellifera L)

Different modes of cell death have been revealed in the regressing hypopharyngeal glands of worker honey bees. The hypopharyngeal gland, which is well developed in young nursing bees to produce protein for larval food, was seen to regress naturally in foraging adult worker bees. A range of techniques including histology, cytochemistry, in situ TUNEL, Annexin V and Comet assays indicated that cells within the gland demonstrate progressive symptoms of apoptosis, necrosis and a vacuolar form of programmed cell death. The latter mode of cell death did not display chromatin margination, but was accompanied by an enhanced level of autophagic and hydrolytic activity in which a cytosolic source of acid phosphatase became manifest in the extra-cisternal spaces. Normal and annexin-positive cells were found to occur in the younger nursing bees, whilst necrosis and an aberrant vacuolar type of apoptosis predominated in the older foraging bees. The relevance of these results to the classification of programmed cell death is discussed. (C) 2000 Academic Press.

On the spectrum of stochastic perturbations of the shift and Julia sets

Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)

Neural networks in chemosystematic studies of asteraceae: A classification based on a dichotomic approach

This paper describes the application of artificial neural nets as an alternative and efficient method for the classification of botanical taxa based on chemical data (chemosystematics). A total of 28,000 botanical occurrences of chemical compounds isolated from the Asteraceae family were chosen from the literature, and grouped by chemical class for each species. Four tests were carried out to differentiate and classify different botanical taxa. The qualifying capacity of the artificial neural nets was dichotomically tested at different hierarchical levels of the family, such as subfamilies and groups of Heliantheae subtribes. Furthermore, two specific subtribes of the Heliantheae and two genera of one of these subtribes were also tested. In general, the artificial neural net gave rise to good results, with multiple-correlation values R > 0.90. Hence, it was possible to differentiate the dichotomic character of the botanical taxa studied.

Submucous myomas: A new presurgical classification to evaluate the viability of hysteroscopic surgical treatment - Preliminary report

STUDY OBJECTIVE: To develop a new preoperative classification of submucous myomas for evaluating the viability and the degree of difficulty of hysteroscopic myomectomy.DESIGN: Retrospective study (Canadian Task Force classification II-3)SETTING: University teaching hospitals.PATIENTS: Fifty-five patients who underwent hysteroscopic resection of submucous myomas.INTERVENTION: the possibility of total resection of the myoma, the operating time, the fluid deficit, and the frequency of any complications were considered. The myomas were classified according to the Classification of the European Society for Gynaecological Endoscopy (ESGE) and by our group's new classification (NC), which considers not only the degree of penetration of the myoma into the myometrium, but also adds in such parameters as the distance of the base of the myoma from the uterine wall, the size of the nodule (cm), and the topography of the uterine cavity. The Fisher's exact test, the Student's t test, and the analysis of variance test were used in the statistical analysis. A p value less than .05 in the two-tailed test was considered significant.MEASUREMENTS AND MAIN RESULTS: In 57 myomas, hysteroscopic surgery was considered complete. There was no significant difference among the three ESGE levels (0, 1, and 2). Using the NC, the difference between the numbers of complete surgeries was significant (p < .001) for the two levels (groups I and H). The difference between the operating times was significant for the two classifications. With respect to the fluid deficit, only the NC showed significant differences between the levels (p = .02).CONCLUSIONS: We believe that the NC gives more clues as to the difficulties of a hysteroscopic myomectomy than the standard ESGE classification. It should be stressed that the number of hysteroscopic myomectomies used in this analysis was low, and it would be interesting to evaluate the performance of the classification in a larger number of patients. (c) 2005 AAGL. All rights reserved.

Estimation of growth fraction in fine needle aspirates from non-Hodgkin's lymphoma using anti-proliferating cell nuclear antigen - Correlation with the Kiel classification

OBJECTIVE: This study was undertaken to assess whether fine needle aspirates from non-Hodgkins lymphoma (NHL) could be used for growth fraction analysis with proliferating cell nuclear antigen (PCNA) staining and if there was a relationship between the growth fraction and cytomorphologic classification according to the Kiel classification.STUDY DESIGN: the study group consisted of 40 patients with NHL diagnosed by fine needle aspiration (FNA) cytology. The cytologic classification of the lymphomas was made by two cytopathologists on May-Grunwald-Giemsa-stained slides using the Kiel classification. There were 27 cases of low and 13 of high grade lymphoma. The estimation of the growth fraction was made by PCNA immunoreactivity. The PCNA index was quantitated in smears by counting an average of 1,000 cells, and the count teas correlated with the cytomorphologic classification.RESULTS: There was It strong correlation between the PCNA index and lymphoma grading. High grade lymphomas exhibited a mean PCNA positivity of 74.0%, which was significantly higher (P <.001) than that of low grade lymphomas (17.6%).CONCLUSION: Our study showed that PCNA evalua tion is suitable for smears obtained by FNA on NHL, correlates with increasing grades of lymphoma according to the Kiel classification and may offer a method of monitoring treatment of lymphoma.