SEARCH · Search PubMed
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.
An analog neural network for support vector machine learning is proposed, based on a partially dual formulation of the quadratic programming problem. It results in a simpler circuit implementation with respect to existing neural solutions for the same application. The effectiveness of the proposed network is shown through some computer simulations concerning benchmark problems.
Kernel methods, such as the support vector machine (SVM), are often formulated as quadratic programming (QP) problems. However, given m training patterns, a naive implementation of the QP solver takes O(m3) training time and at least O(m2) space. Hence, scaling up these QPs is a major stumbling block in applying kernel methods on very large data sets, and a replacement of the naive method for finding the QP solutions is highly desirable. Recently, by using approximation algorithms for the minimum enclosing ball (MEB) problem, we proposed the core vector machine (CVM) algorithm that is much faster and can handle much larger data sets than existing SVM implementations. However, the CVM can only be used with certain kernel functions and kernel methods. For example, the very popular support vector regression (SVR) cannot be used with the CVM. In this paper, we introduce the center-constrained MEB problem and subsequently extend the CVM algorithm. The generalized CVM algorithm can now be used with any linear/nonlinear kernel and can also be applied to kernel methods such as SVR and the ranking SVM. Moreover, like the original CVM, its asymptotic time complexity is again linear in m and its space complexity is independent of m. Experiments show that the generalized CVM has comparable performance with state-of-the-art SVM and SVR implementations, but is faster and produces fewer support vectors on very large data sets.
A novel learning algorithm, the Recurrent Neural Network Constrained Optimization Method (RENNCOM) is suggested in this paper, for training block-diagonal recurrent neural networks. The training task is formulated as a constrained optimization problem, whose objective is twofold: (1) minimization of an error measure, leading to successful approximation of the input/output mapping and (2) optimization of an additional functional, the payoff function, which aims at ensuring network stability throughout the learning process. Having assured the network and training stability conditions, the payoff function is switched to an alternative form with the scope to accelerate learning. Simulation results on a benchmark identification problem demonstrate that, compared to other learning schemes with stabilizing attributes, the RENNCOM algorithm has enhanced qualities, including, improved speed of convergence, accuracy and robustness. The proposed algorithm is also applied to the problem of the analysis of lung sounds. Particularly, a filter based on block-diagonal recurrent neural networks is developed, trained with the RENNCOM method. Extensive experimental results are given and performance comparisons with a series of other models are conducted, underlining the effectiveness of the proposed filter.
A nonparametric Bayesian formulation is given to the problem of modeling nonhomogeneous spatial point patterns influenced by concomitant variables. Only incomplete information on the concomitant variables is assumed, consisting of a relatively small number of point measurements. Residual variation, caused by other unmeasured influential factors, is modeled in terms of a spatially varying baseline intensity function. A Markov chain Monte Carlo scheme is proposed for the simultaneous nonparametric estimation of each unknown function in the model. The suggested method is illustrated by reanalysing a data set in Rathbun (1996, Biometrics 52, 226-242), and the estimated models are compared with those obtained by Rathbun.
According to the Cognitive Complexity and Control (CCC) theory, the development of executive function can be understood in terms of age-related increases in the maximum complexity of the rules children can formulate and use when solving problems. This Monograph describes four studies (9 experiments) designed to test hypotheses derived from the CCC theory and from alternative theoretical perspectives on the development of executive function (memory accounts, inhibition accounts, and redescription accounts). Each study employed a version of the Dimensional Change Card Sort (DCCS), in which children are required first to sort cards by one pair of rules (e.g., color rules: "If red then here, if blue then there"), and then sort the same cards by another, incompatible pair of rules (e.g., shape rules). Study 1 found that although most 3- to 4-year-olds failed the standard version of this task (i.e., they perseverated on the preswitch rules during the postswitch phase), they usually performed well when they were required to use four rules (including bidimensional rules) and those rules were not in conflict (i.e., they did not require children to respond in two different ways to the same test card). These findings indicate that children's perseveration cannot be attributed in a straightforward fashion to limitations in children's memory capacity. Study 2 examined the circumstances in which children can use conflicting rules. Three experiments demonstrated effects of rule dimensionality (uni- vs. bidimensional rules) but no effects of stimulus characteristics (1 vs. 2 test cards; spatially integrated vs. separated stimuli). Taken together, these studies suggest that conflict among rules is a key determinant of difficulty, but that conflict interacts with dimensionality. Study 3 examined what types of conflict pose problems for 3- to 4-year-olds by comparing performance on standard, Partial Change, and Total Change versions of the DCCS. Results revealed effects of conflict at the level of specific rules (e.g., "If red, then there"), rather than specific stimulus configurations or dimensions per se, indicating that activation of the preswitch rules persists into the postswitch phase. Study 4 examined whether negative priming also contributes to difficulty on the DCCS. Two experiments suggested that the active selection of preswitch rules against a competing alternative results in the lasting suppression of the alternative. Taken together, the results of these studies provide the basis for a revision of the CCC theory (CCC-r) that specifies more clearly the circumstances in which children will have difficulty using rules at various levels of complexity, provides a more detailed account of how to determine the complexity of rules required in a task, takes account of both the activation and inhibition of rules as a function of experience, and highlights the importance of taking intentionality seriously in the study of executive function.
Experimental approaches to modelling the enzymatic function of biological membranes are discussed. Emphasis is given to pseudohomogeneous systems such as proteolipid complexes and enzymes in organic solvents; the latter are solubilized with phospholipids or synthetic surfactants. Methods for producing and studying such micellar systems are considered. The key research problems of micellar enzymology are formulated and its relation to enzyme membranology is discussed. Finally, the new potentialities are noted of applied enzymology (biotechnology) offered by application of a colloidal solution of water in organic solvents as a microheterogeneous medium for enzymatic reactions.
The introduction of cyclosporine (CsA) in clinical practice has significantly improved patient and allograft survival after organ transplantation. The new microemulsion CsA formulation, Neoral, has been associated with a more reproducible absorption and a better patient outcome as compared to the old formulation Sandimmune. Recently, several generic CsA formulations have been tested as bioequivalent to Neoral. Bioequivalence tests have been performed in selected groups of young, healthy male volunteers usually in single-dose studies, and then extended to completely different population, such as transplant recipients. However, growing body of evidence shows that CsA pharmacokinetics in healthy subjects is different from that of transplant patients, treated chronically with CsA. Therefore, converting patients from Neoral to the new generic formulations could be detrimental, exposing patients to increased risk of graft function deterioration and graft loss. Thus, more research and more accurate bioequivalence tests are required to address the unanswered problems dealing with the generic CsA formulations.
When approaching the study of terminal growth and segmentation, comparative morphology provides an important guide to formulate questions. There are often problems in unambiguously identifying the axis along which we wish to study terminal (often, actually, subterminal) growth, especially when the trunk axis is posteriorly prolonged in an appendage (as with the tail of vertebrates), or when the polarity of the "external animal" is other than the polarity of the "internal animal," as in polypoid bilaterians. We cannot ignore that the rear end of the main body axis is possibly defined very early in development in some groups, for example, arthropods, whereas in others, vertebrates, for example, it is defined much later. We cannot think of segmentation as always corresponding to the sequential posterior addition of new units, thus ignoring the widespread occurrence of double segmentation. A more subtle problem is represented by the overlapping of different processes, all of them contributing to elongating the body, such as segmentation, cell proliferation, cell rearrangement, and cell growth. Within a segmented trunk, cell proliferation and differentiation may go on in parallel from as many growth points as there are groups of regularly spaced cells. The main consequence, however, is not so much to expedite elongation as to reduce the disparity of metabolic conditions, gene expression patterns and "relative age" of different body districts, otherwise possibly troublesome within the limited space of the embryo.
The fundamental problems of schizophrenia are wrongly formulated unless one takes heed of the major philosophical revolution which Kant initiated with the publication of the Critique of Pure Reason in 1781. Virtually all major philosophers since then have adhered to the view that the mind objectifies whatever energy fields/atoms, etc., are 'out there'. The thesis put forward in this lecture will be that schizophrenia is a condition in which this process proceeds too far. Evidence from the psychopathology of body, time and world is offered in support of such morbid objectivization.
The fields of gerontology and genetics have merged, spawning novel lines of investigation and generating a wealth of new results in recent years. However, the lack of clarity and consistency in the basic definitions upon which the science of gerontology must rest has fostered a certain amount of enduring confusion. Among the unclear issues are the genetic components of life span and the distinction between "normal" and "pathologic" aging. At a time of massive world population aging, such issues have, beyond their scientific importance, a momentous social and economic impact. A simple axiomatic framework, consisting of three definitions and five axioms, is proposed that clarifies the aforementioned issues and reconciles disparate data in gerontology. Based on this framework, a new classification of genes involved in survival is proposed. Within the Compensatory Adaptation Theory of aging, apparent paradoxes are solved and problems in gerontology may be formulated anew.
This paper includes formulation of the multiconnected transient problem for an underwater transducer system of two thin-walled cylindrical piezoelectric ceramic radiators, description of an analytical method of solution, and investigation of fundamental regularities of transient processes. All these points take into account the interaction of reflectors. Radiators with continuous and sectioned electrodes are considered during excitation with different electrical impulses. An experimental test of selected mathematical models has been done.
The principal conclusion to be drawn from the foregoing discussion is that the death of cells and the destruction of tissues, organs, and organ systems are programmed as normal morphogenetic events in the development of multicellular organisms. Death in embryonic systems may thus be explored within the same conceptual framework as growth and differentiation. The present exploration has revealed that death during embryogenesis serves utilitarian goals in some instances, at least, that its occurrence is subject to control by factors of the immediate cellular and humoral environment, and that aberrations in its normal pattern of expression provide the mechanism for realization of many mutant phenotypes. Hopefully, it has also pointed toward the appropriate formulation of some of the problems that confront us in understanding the control of death at the level of genetic transcription, the biochemical events which determine and accompany its occurrence, and the pathways of disposition and the developmental significance of disassembled cellular building blocks.
Most regional health authorities throughout the United Kingdom have established drug information units to provide health service staff with a wide range of information about drugs and drug use. The units, which are staffed by drug information pharmacists, provide their service mainly by answering inquiries, although some disseminate information more positively through lectures and bulletins.An analysis of inquiries received by regional information units during 1976 showed that most were submitted by hospital doctors or pharmacists; comparatively few were received from general practitioners. Topics of inquiry included adverse effects of drugs, source of supply and identification, current treatment, dosage, route, precautions, and pharmaceutical problems such as stability or formulation of drug preparations. A more detailed analysis of the inquiries received by the North-western Regional Drug Information Service at Manchester over three years showed that the number of inquiries gradually increased and that more were received from general practitioners after a programme of lectures had been introduced to tell them about the service. The North-western service also received more requests from hospital pharmacists than other units, though many originated from clinicians.The regional drug information units consulted widely with clinical and other specialists in answering questions, but about a quarter of all inquiries were pharmaceutical, relating to stability and incompatibility. A multidisciplinary approach therefore seems necessary to provide a comprehensive and advisory drug information service.
During orthostatic stress, arterial and cardiopulmonary baroreflexes play a key role in maintaining arterial pressure by regulating heart rate. This study presents a mathematical model that can predict the dynamics of heart rate regulation in response to postural change from sitting to standing. The model uses blood pressure measured in the finger as an input to model heart rate dynamics in response to changes in baroreceptor nerve firing rate, sympathetic and parasympathetic responses, vestibulo-sympathetic reflex, and concentrations of norepinephrine and acetylcholine. We formulate an inverse least squares problem for parameter estimation and successfully demonstrate that our mathematical model can accurately predict heart rate dynamics observed in data obtained from healthy young, healthy elderly, and hypertensive elderly subjects. One of our key findings indicates that, to successfully validate our model against clinical data, it is necessary to include the vestibulo-sympathetic reflex. Furthermore, our model reveals that the transfer between the nerve firing and blood pressure is nonlinear and follows a hysteresis curve. In healthy young people, the hysteresis loop is wide, whereas, in healthy and hypertensive elderly people, the hysteresis loop shifts to higher blood pressure values, and its area is diminished. Finally, for hypertensive elderly people, the hysteresis loop is generally not closed, indicating that, during postural change from sitting to standing, baroreflex modulation does not return to steady state during the first minute of standing.
This article is a discussion of the consequences of "the problem of induction" as manifested in clinical decision making. All data, statistics, and probabilities are concerned with groups, generalizing from samples to other populations. It is not self-evident what such processes mean at the level of the individual patient. Some discussion of this problem is followed by a formulation of subjective probability that may contribute to the application and acceptance of clinical decision analysis.
The authors investigated strategies employed by resident physicians to decide whether to prescribe hormonal replacement therapy (HRT) for menopausal women, a matter of continuing clinical controversy. Verbal protocols were obtained from 21 residents in three specialties as they responded to 12 brief case descriptions. The cases incorporated three levels of cancer risk and two levels of osteoporosis risk in a 3 x 2 factorial design with two replications in each cell. Substantial variation in willingness to prescribe HRT was observed. By clustering subjects with relatively similar approaches to the problem, three treatment strategies were formulated that accounted for the decisions of 20 subjects. Each strategy is a simplified representation of the conflicting considerations in this clinical dilemma that facilitates rapid decision making. The differences between these representations and formal decision-analytic models help to explain why observed clinical decisions were inconsistent with expected utility maximization.
The complex pharmacokinetics that promote transit of a drug from ingestion to elimination are characterized by variability. The rate and degree to which a drug undergoes absorption, distribution, biotransformation and excretion are a result of both endogenous and exogenous factors, some of which can be predicted, many of which cannot. One of the greatest problems associated with conventional drug formulations has been the dose-related appearance of side-effects that often accompany the initial surge of drug release. Drugs with short half-lives or narrow therapeutic ratios may alternately produce over-medication and under-medication in a patient. Innovations in drug form design have resulted in a new precision-release formulation that permits zero-order steady state drug release. Sodium indomethacin trihydrate ('Osmosin') is the first drug to be contained within this new precision-release formulation which approaches the objective of delivering drugs at a therapeutically optimal rate.
BACKGROUND: Single nucleotide polymorphisms (SNPs) are locations at which the genomic sequences of population members differ. Since these differences are known to follow patterns, disease association studies are facilitated by identifying SNPs that allow the unique identification of such patterns. This process, known as haplotype tagging, is formulated as a combinatorial optimization problem and analyzed in terms of complexity and approximation properties. RESULTS: It is shown that the tagging problem is NP-hard but approximable within 1 + ln((n2 - n)/2) for n haplotypes but not approximable within (1-epsilon) ln(n/2) for any epsilon > 0 unless NP subset DTIME(n(log log n)). A simple, very easily implementable algorithm that exhibits the above upper bound on solution quality is presented. This algorithm has running time O(np/2(2m-p+1)) < or = O(m(n2-n)/2) where p < or = min(n, m) for n haplotypes of size m. As we show that the approximation bound is asymptotically tight, the algorithm presented is optimal with respect to this asymptotic bound. CONCLUSION: The haplotype tagging problem is hard, but approachable with a fast, practical, and surprisingly simple algorithm that cannot be significantly improved upon on a single processor machine. Hence, significant improvement in computational efforts expended can only be expected if the computational effort is distributed and done in parallel.