Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Networks”

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 631 records · Page 35Linked to original sources
PubMed1996

Radial Basis Function Network Configuration Using Mutual Information and the Orthogonal Least Squares Algorithm.

Input nodes of neural networks are usually predetermined by using a priori knowledge or selected by trial and error. For example, in pattern recognition applications the input nodes are usually the given pattern features and in system identification applications the past input and output data are often used as inputs to the network. Some of the input variables may be irrelevant to the task in hand and therefore may cause a deterioration in network performance. Some may be redundant and may increase the complexity of the network and consume expensive computation time. In the present study, the mutual information between the input variables and the output of the network is used to select a suboptimal set of input variables for the network. The variables are selected according to the information content relevant to the output. Variables which have a higher mutual information with the output and lower dependence on other selected variables are used as network inputs. The algorithms are derived based on heuristics and performance is assessed by using radial basis function (RBF) networks trained with the orthogonal least squares algorithm (OLS), which selects the hidden layer nodes of the network according to the error reduction ratios on the network output. Both real and simulated data sets are used to demonstrate the effectiveness of the new algorithms. Copyright 1996 Elsevier Science Ltd.

Journal Article↗
PubMed1996

On Temporal Generalization of Simple Recurrent Networks.

Simple recurrent networks (Elman networks) have been widely used in temporal processing applications. In this study we investigate temporal generalization of simple recurrent networks, drawing comparisons between network capabilities and human performance. Elman networks are trained to generate temporal trajectories sampled at different rates. The networks are then tested with trajectories at the trained rates and other sampling rates, including trajectories representing mixtures of different sampling rates. It is found that for simple trajectories the networks show interval invariance, but not rate invariance. However, for complex trajectories which require greater contextural information, these networks do not seem to show any temporal generalization. Similar results are also obtained using measured speech data. These results suggest that this class of recurrent networks exhibits severe limitations in temporal generalization. Discussions are provided regarding rate invariance and possible ways to achieve it in neural networks. Copyright 1996 Elsevier Science Ltd

Journal Article↗
PubMed2026

Cells and Networks in Flux: Rethinking Ontogenesis and Pathogenesis.

Organ and tissue functions emerge from the coordinated activity of cell networks. Therapeutics that act on pathogenic cell networks, modulating their cellular interplay, follow naturally. Over several decades, our laboratory has developed a series of approaches for rewiring cell networks, culminating in a class of cell surface-directed signal converter proteins (SCPs) that do so by modulating juxtacrine and autocrine signaling in and among their nodal cells. A first such SCP has now produced encouraging clinical data for cancer immunotherapy. Yet, these early network-directed fusion proteins rest on a deliberately simplified picture: discrete end-cell types plugged into graphically tractable networks. That picture is increasingly at odds with what computational cell typing and spatiotemporal analytics, along with epigenetics, now reveal-a hyperdiverse, plastic, experience-shaped cellular landscape embedded in dynamic, multiway networks. Setting the stage for a next generation of network modulators, an extended cell differentiation synthesis is proposed, which formalizes paracell ultradifferentiation and aging-associated differentiation phases. According to this model, cells are ever evolving, and no two cells are alike. A richer cellular ontology forces a more elaborate network ontology, with paralogous networks and their shifting subnetworks opening a concrete design space for next-generation network-directed SCP therapeutics. This exploration calls for a willingness to embrace complexity more fully and borrow freely from conceptual fields close and afar.

Humans↗
PubMed2003

Maturation of rhythmic neural network: role of central modulatory inputs.

Modulatory systems are well known for their roles in tuning the cellular and synaptic properties in the adult neuronal networks, and play a major role in the control of the flexibility of functional outputs. However far less is known concerning their role in the maturation of neural networks during the development. In this review, using the stomatogastric nervous system of lobster, we will show that the neuromodulatory system exerts a powerful influence on developing neural networks. In the adult the number of both motor target neurons and their modulatory neurons is restricted to tens of identifiable cells. They are therefore well characterized in terms of cellular, synaptic and morphological properties. In the embryo, these target cells and their neuromodulatory population are already present from mid-embryonic life. However, the motor output generated by the system is quite different: while in the embryo all the target neurons are organized into a single network generating unique motor pattern, in the adult this population splits into two distinct networks generating separate patterns. This ontogenetic partitioning does not rely on progressive acquisition of adult properties but rather on a switch between two possible network operations. Indeed, adult networks are present early in the embryonic life but their expression is repressed by central modulatory neurons. Moreover, embryonic networks can be revealed in the adult system again by altering modulatory influences. Therefore, independently of the developmental age, two potential network phenotypes co-exist within the same neuronal architecture: when one is expressed, the other one is hidden and vice versa. These transitions do not necessarily need dramatic changes such as growth/retraction of processes, acquisition of new intra-membrane proteins etc. but rather, as shown by modelling studies, it may simply rely on a subtle tuning of pre-existing intercellular electrical coupling. This in turn suggests that progressive ontogenetic alteration may not take place at the level of the target network but rather at the level of modulatory input neurons.

Action Potentials↗
PubMed2005

The smallest of all worlds: pollination networks.

A pollination network may be either 2-mode, describing trophic and reproductive interactions between communities of flowering plants and pollinator species within a well-defined habitat, or 1-mode, describing interactions between either plants or pollinators. In a 1-mode pollinator network, two pollinator species are linked to each other if they both visit the same plant species, and vice versa for plants. Properties of 2-mode networks and their derived 1-mode networks are highly correlated and so are properties of 1-mode pollinator and 1-mode plant networks. Most network properties are scale-dependent, i.e. they are dependent upon network size. Pollination networks have the strongest small-world properties of any networks yet studied, i.e. all species are close to each other (short average path length) and species are highly clustered. Species in pollination networks are much more densely linked than species in traditional food webs, i.e. they have a higher density of links, a shorter distance between species, and species are more clustered.

Animals↗
PubMed2006

Exploring local structural organization of metabolic networks using subgraph patterns.

Metabolic networks of many cellular organisms share global statistical features. Their connectivity distributions follow the long-tailed power law and show the small-world property. In addition, their modular structures are organized in a hierarchical manner. Although the global topological organization of metabolic networks is well understood, their local structural organization is still not clear. Investigating local properties of metabolic networks is necessary to understand the nature of metabolism in living organisms. To identify the local structural organization of metabolic networks, we analysed the subgraphs of metabolic networks of 43 organisms from three domains of life. We first identified the network motifs of metabolic networks and identified the statistically significant subgraph patterns. We then compared metabolic networks from different domains and found that they have similar local structures and that the local structure of each metabolic network has its own taxonomical meaning. Organisms closer in taxonomy showed similar local structures. In addition, the common substrates of 43 metabolic networks were not randomly distributed, but were more likely to be constituents of cohesive subgraph patterns.

Algorithms↗
PubMed2006

Transient rhythmic network activity in the somatosensory cortex evoked by distributed input in vitro.

The initiation and maintenance of physiological and pathophysiological oscillatory activity depends on the synaptic interactions within neuronal networks. We studied the mechanisms underlying evoked transient network oscillation in acute slices of the adolescent rat somatosensory cortex and modeled its underpinning mechanisms. Oscillations were evoked by brief spatially distributed noisy extracellular stimulation, delivered via bipolar electrodes. Evoked transient network oscillation was detected with multi-neuron patch-clamp recordings under different pharmacological conditions. The observed oscillations are in the frequency range of 2-5 Hz and consist of 4-12 mV large, 40-150 ms wide compound synaptic events with rare overlying action potentials. This evoked transient network oscillation is only weakly expressed in the somatosensory cortex and requires increased [K+]o of 6.25 mM and decreased [Ca2+]o of 1.5 mM and [Mg2+]o of 0.5 mM. A peak in the cross-correlation among membrane potential in layers II/III, IV and V neurons reflects the underlying network-driven basis of the evoked transient network oscillation. The initiation of the evoked transient network oscillation is accompanied by an increased [K+]o and can be prevented by the K+ channel blocker quinidine. In addition, a shift of the chloride reversal potential takes place during stimulation, resulting in a depolarizing type A GABA (GABAA) receptor response. Blockade of alpha-amino-3-hydroxy-5-methyl-4-isoxazole-proprionate (AMPA), N-methyl-D-aspartate (NMDA), or GABA(A) receptors as well as gap junctions prevents evoked transient network oscillation while a reduction of AMPA or GABA(A) receptor desensitization increases its duration and amplitude. The apparent reversal potential of -27 mV of the evoked transient network oscillation, its pharmacological profile, as well as the modeling results suggest a mixed contribution of glutamatergic, excitatory GABAergic, and gap junctional conductances in initiation and maintenance of this oscillatory activity. With these properties, evoked transient network oscillation resembles epileptic afterdischarges more than any other form of physiological or pathophysiological neocortical oscillatory activity.

Animals↗
PubMed2006

Gene interaction network analysis suggests differences between high and low doses of acetaminophen.

Bayesian networks for quantifying linkages between genes were applied to detect differences in gene expression interaction networks between multiple doses of acetaminophen at multiple time points. Seventeen (17) genes were selected from the gene expression profiles from livers of rats orally exposed to 50, 150 and 1500 mg/kg acetaminophen (APAP) at 6, 24 and 48 h after exposure using a variety of statistical and bioinformatics approaches. The selected genes are related to three biological categories: apoptosis, oxidative stress and other. Gene interaction networks between all 17 genes were identified for the nine dose-time observation points by the TAO-Gen algorithm. Using k-means clustering analysis, the estimated nine networks could be clustered into two consensus networks, the first consisting of the low and middle dose groups, and the second consisting of the high dose. The analysis suggests that the networks could be segregated by doses and were consistent in structure over time of observation within grouped doses. The consensus networks were quantified to calculate the probability distribution for the strength of the linkage between genes connected in the networks. The quantifying analysis showed that, at lower doses, the genes related to the oxidative stress signaling pathway did not interact with the apoptosis-related genes. In contrast, the high-dose network demonstrated significant interactions between the oxidative stress genes and the apoptosis genes and also demonstrated a different network between genes in the oxidative stress pathway. The approaches shown here could provide predictive information to understand high- versus low-dose mechanisms of toxicity.

Acetaminophen↗
PubMed2003

Identification of the requisite brain sites in the neuronal network subserving generalized clonic audiogenic seizures.

Comparative studies of neuronal networks that subserve convulsions in closely-related epilepsy models are revealing instructive data about the pathophysiological mechanisms that govern these networks. Studies of audiogenic seizures (AGS) in genetically epilepsy-prone rats (GEPRs) and related forms of AGS demonstrate important network similarities and differences. Two substrains of GEPRs exist, GEPR-9s, exhibiting tonic AGS, and GEPR-3s, exhibiting clonic AGS. The neuronal network for tonic AGS resides exclusively in brainstem nuclei, but forebrain sites, including the amygdala (AMG), are recruited after repetitive AGS induction. The neuronal network for clonic AGS remains to be investigated. The present study examined the neuronal network for clonic AGS in GEPR-3s by microinjecting a competitive NMDA receptor antagonist, D,L-2-amino-7-phosphonoheptanoic acid (AP7), into the central nucleus of inferior colliculus (ICc), deep layers of superior colliculus (DLSC), periaqueductal grey (PAG), or caudal pontine reticular formation (cPRF), which are implicated in tonic AGS networks. Microinjections into AMG and perirhinal cortex (PRh), which are not implicated in AGS, were also done. AGS in GEPR-3s were blocked reversibly after microinjections into ICc, DLSC, PAG or cPRF. However, AGS were also blocked by AP7 in AMG but not PRh. The sites in which AP7 blocks AGS are implicated as requisite components of the clonic AGS network, and these data support a critical role for NMDA receptors in clonic AGS modulation. The brainstem nuclei of the clonic AGS network are identical to those subserving tonic AGS. However, the requisite involvement of AMG in the clonic AGS network, which is not seen in tonic AGS, is surprising and suggests important mechanistic differences between clonic and tonic forms of AGS.

Acoustic Stimulation↗
PubMed1999

Possible mechanisms through which dietary pectin influences fibrin network architecture in hypercholesterolaemic subjects.

It is suspected that not only fibrinogen concentration but also the quality of fibrin networks may contribute to cardiovascular risk. Evidence is accumulating that a "prudent" diet may protect against diseases associated with raised clotting factors. The effect of diet on fibrinogen is, however, still controversial. In a previous study performed in our laboratory, it was shown that dietary pectin influences fibrin network architecture in hypercholesterolaemic men without causing any changes in fibrinogen concentration. To elucidate the possible mechanisms, it was necessary to study the possibility that pectin may itself have indirect effects on fibrin network architecture. Pectin is fermented in the gastrointestinal tract to acetate, propionate, and butyrate. In humans, only acetate reaches the circulation beyond the liver. This investigation primarily examined the possibility that pectin may, through acetate, influence fibrin network architecture in vivo. The effects of pectin and acetate supplementation in hypercholesterolaemic subjects were compared. Furthermore, this study also aimed at describing the possible in vitro effects of acetate on fibrin network architecture. Two groups of 10 male hyperlipidaemic volunteers each received a pectin (15 g/day) or acetate (6.8 g/day) supplement for 4 weeks. Acetate supplementation did not cause a significant change in plasma fibrinogen levels. As in the pectin group, significant differences were found in the characteristics of fibrin networks developed in plasma after 4 weeks of acetate supplementation. Fibrin networks were more permeable (from 213+/-76 to 307+/-81 x 10(11) cm2), had lower tensile strength (from 23+/-3 to 32+/-9% compaction), and were more lyseable (from 252+/-11 to 130+/-15 minutes). These results strongly suggest that the effect of pectin on network architecture could partially be mediated by acetate. Progressive amounts of acetate were used in vitro to investigate the possibility that acetate may be directly responsible for changes that occurred in fibrin network architecture in the plasma medium. Results indicated that acetate influenced fibrin network architecture directly. From the results, it seems highly possible that acetate may be responsible in part for the beneficial effects of pectin supplementation in vivo. It is evident that pectin or acetate supplementation can be useful during the treatment or prevention of some clinical manifestations, especially those associated with raised total cholesterol and possibly also plasma fibrinogen.

Adult↗
PubMed2004

The view from two worlds: The convergence of social network reports between mental health clients and their ties.

Traditionally, concerns with the similarities and discrepancies between the reports of persons (or focal respondents) and their collaterals (or network ties/respondents) about the former's social support network is framed as a methodological concern. As individuals experience the devastating effects of illness, and especially as their cognitive capabilities or social perceptions may be impaired by mental health problems, these listings are seen as potentially problematic. While we share this concern, we expand the investigation of the comparison of network ties from focal and network respondents to consider the nature of differences. Using data from Wave I of the Indianapolis Network Mental Health Study, we target the "health matters" network of individuals making their first major contact with the city's largest public and voluntary facilities. Overall, we find that the networks on which "first timers" rely to discuss health matters are small, with both focal and network respondents mentioning four individuals on average. The overlap in ties mentioned is just over 2 people (or 56 percent), on average, and differ with regard to the number of friends and health care professionals mentioned. Ironically, listings are more accurate for focal respondents who have more serious mental illnesses or larger networks. The extent of overlap is lower for women focal respondents than men. In sum, while convergence is in a range considered acceptable in network studies, the substantive nature of discrepancies have interesting and important theoretical and clinical implications.

Adolescent↗
PubMed1999

Hierarchies and cliques in the social networks of health care professionals: implications for the design of dissemination strategies.

Interest in how best to influence the behaviour of clinicians in the interests of both clinical and cost effectiveness has rekindled concern with the social networks of health care professionals. Ever since the seminal work of Coleman et al. [Coleman, J.S., Katz, E., Menzel, H., 1966. Medical Innovation: A Diffusion Study. Bobbs-Merrill, Indianapolis.], networks have been seen as important in the process by which clinicians adopt (or fail to adopt) new innovations in clinical practice. Yet very little is actually known about the social networks of clinicians in modern health care settings. This paper describes the professional social networks of two groups of health care professionals, clinical directors of medicine and directors of nursing, in hospitals in England. We focus on network density, centrality and centralisation because these characteristics have been linked to access to information, social influence and social control processes. The results show that directors of nursing are more central to their networks than clinical directors of medicine and that their networks are more hierarchical. Clinical directors of medicine tend to be embedded in much more densely connected networks which we describe as cliques. The hypotheses that the networks of directors of nursing are better adapted to gathering and disseminating information than clinical directors of medicine, but that the latter could be more potent instruments for changing, or resisting changes, in clinical behaviour, follow from a number of sociological theories. We conclude that professional socialisation and structural location are important determinants of social networks and that these factors could usefully be considered in the design of strategies to inform and influence clinicians.

Adult↗
PubMedDate not supplied

Recurrent fractal neural networks: a strategy for the exchange of local and global information processing in the brain.

The regulation of biological networks relies significantly on convergent feedback signaling loops that render a global output locally accessible. Ideally, the recurrent connectivity within these systems is self-organized by a time-dependent phase-locking mechanism. This study analyzes recurrent fractal neural networks (RFNNs), which utilize a self-similar or fractal branching structure of dendrites and downstream networks for phase-locking of reciprocal feedback loops: output from outer branch nodes of the network tree enters inner branch nodes of the dendritic tree in single neurons. This structural organization enables RFNNs to amplify re-entrant input by over-the-threshold signal summation from feedback loops with equivalent signal traveling times. The columnar organization of pyramidal neurons in the neocortical layers V and III is discussed as the structural substrate for this network architecture. RFNNs self-organize spike trains and render the entire neural network output accessible to the dendritic tree of each neuron within this network. As the result of a contraction mapping operation, the local dendritic input pattern contains a downscaled version of the network output coding structure. RFNNs perform robust, fractal data compression, thus coping with a limited number of feedback loops for signal transport in convergent neural networks. This property is discussed as a significant step toward the solution of a fundamental problem in neuroscience: how is neuronal computation in separate neurons and remote brain areas unified as an instance of experience in consciousness? RFNNs are promising candidates for engaging neural networks into a coherent activity and provide a strategy for the exchange of global and local information processing in the human brain, thereby ensuring the completeness of a transformation from neuronal computation into conscious experience.

Brain↗
PubMed1997

Optimal Linear Combinations of Neural Networks.

Neural network-based modeling often involves trying multiple networks with different architectures and training parameters in order to achieve acceptable model accuracy. Typically, one of the trained networks is chosen as best, while the rest are discarded. [Hashem and Schmeiser (1995)] proposed using optimal linear combinations of a number of trained neural networks instead of using a single best network. Combining the trained networks may help integrate the knowledge acquired by the components networks and thus improve model accuracy. In this paper, we extend the idea of optimal linear combinations (OLCs) of neural networks and discuss issues related to the generalization ability of the combined model. We then present two algorithms for selecting the component networks for the combination to improve the generalization ability of OLCs. Our experimental results demonstrate significant improvements in model accuracy, as a result of using OLCs, compared to using the apparent best network. Copyright 1997 Elsevier Science Ltd.

Journal Article↗
PubMed2006

Effects of the implementation of a telemedical stroke network: the Telemedic Pilot Project for Integrative Stroke Care (TEMPiS) in Bavaria, Germany.

BACKGROUND: Telemedical networks are a new approach to improve stroke care in community settings. We aimed to assess the effects of a stroke network with telemedical support in Germany on quality of care, according to acute processes and long-term outcome. METHODS: Five community hospitals without pre-existing specialised stroke care were included in a network with telemedical support by two academic hospitals. In a non-randomised, open intervention study, five community hospitals without specialised stroke care served as the control group, matched individually to the network hospitals by predefined characteristics. Stroke patients admitted consecutively to one of the participating hospitals between July 7, 2003, and March 31, 2005, were included in the study. Patients in network and control hospitals were assessed in the same manner and were followed up for vital status, living situation, and disability at 3 months. Poor outcome was defined by death, institutional care, or disability (Barthel index <60 or modified Rankin scale >3). Predefined indicators for quality of acute stroke care were achieved. FINDINGS: A total of 5696 patients with a sudden, non-convulsive loss of neurological function who were diagnosed with having suspected stroke were admitted to the ten hospitals participating in the study. After exclusion, 3122 were included in the final analysis, of whom 1971 (63%) were treated in the network hospitals. All indicators related to quality of acute stroke care were more commonly met in the network than in the control hospitals. After 3 months, 44% of patients treated in network hospitals versus 54% treated in control hospitals had a poor outcome (p<0.0001). In multivariate regression analysis, treatment in network hospitals independently reduced the probability of a poor outcome (odds ratio 0.62, 95% CI 0.52-0.74; p<0.0001). INTERPRETATION: Telemedical networks with academic stroke centres offer new and innovative approaches to improve acute stroke care at community level for stroke patients living in non-urban areas.

Academic Medical Centers↗
PubMed2006

Two-dimensional porous molecular networks of dehydrobenzo[12]annulene derivatives via alkyl chain interdigitation.

The self-assembly of a series of hexadehydrotribenzo[12]annulene (DBA) derivatives has been scrutinized by scanning tunneling microscopy (STM) at the liquid-solid interface. First, the influence of core symmetry on the network structure was investigated by comparing the two-dimensional (2D) ordering of rhombic bisDBA 1a and triangular DBA 2a (Figure 1). BisDBA 1a forms a Kagomé network upon physisorption from 1,2,4-trichlorobenzene (TCB) onto highly oriented pyrolytic graphite (HOPG). Under similar experimental conditions, DBA 2a shows the formation of a honeycomb network. The core symmetry and location of alkyl substituents determine the network structure. The most remarkable feature of the DBA networks is the interdigitation of the nonpolar alkyl chains: they connect the pi-conjugated cores and direct their orientation. As a result, 2D open networks with voids are formed. Second, the effect of alkyl chain length on the structure of DBA patterns was investigated. Upon increasing the length of the alkyl chains (DBAs 3c-e) a transition from honeycomb networks to linear networks was observed in TCB, an observation attributed to stronger molecule-substrate interactions. Third, the effect of solvent on the structure of the nonpolar DBA networks was investigated in four different solvents: TCB as a polar aromatic solvent, 1-phenyloctane as a solvent having both aromatic and aliphatic moieties, n-tetradecane as an aliphatic solvent, and octanoic acid as a polar alkylated solvent. The solvent dramatically changes the structure of the DBA networks. The solvent effects are discussed in terms of factors that influence the mobility of molecules at the liquid-solid interface such as solvation.

Alkylation↗
PubMed1998

Collective dynamics of 'small-world' networks.

Networks of coupled dynamical systems have been used to model biological oscillators, Josephson junction arrays, excitable media, neural networks, spatial games, genetic control networks and many other self-organizing systems. Ordinarily, the connection topology is assumed to be either completely regular or completely random. But many biological, technological and social networks lie somewhere between these two extremes. Here we explore simple models of networks that can be tuned through this middle ground: regular networks 'rewired' to introduce increasing amounts of disorder. We find that these systems can be highly clustered, like regular lattices, yet have small characteristic path lengths, like random graphs. We call them 'small-world' networks, by analogy with the small-world phenomenon (popularly known as six degrees of separation. The neural network of the worm Caenorhabditis elegans, the power grid of the western United States, and the collaboration graph of film actors are shown to be small-world networks. Models of dynamical systems with small-world coupling display enhanced signal-propagation speed, computational power, and synchronizability. In particular, infectious diseases spread more easily in small-world networks than in regular lattices.

Animals↗
PubMed2000

Error and attack tolerance of complex networks

Many complex systems display a surprising degree of tolerance against errors. For example, relatively simple organisms grow, persist and reproduce despite drastic pharmaceutical or environmental interventions, an error tolerance attributed to the robustness of the underlying metabolic network. Complex communication networks display a surprising degree of robustness: although key components regularly malfunction, local failures rarely lead to the loss of the global information-carrying ability of the network. The stability of these and other complex systems is often attributed to the redundant wiring of the functional web defined by the systems' components. Here we demonstrate that error tolerance is not shared by all redundant systems: it is displayed only by a class of inhomogeneously wired networks, called scale-free networks, which include the World-Wide Web, the Internet, social networks and cells. We find that such networks display an unexpected degree of robustness, the ability of their nodes to communicate being unaffected even by unrealistically high failure rates. However, error tolerance comes at a high price in that these networks are extremely vulnerable to attacks (that is, to the selection and removal of a few nodes that play a vital role in maintaining the network's connectivity). Such error tolerance and attack vulnerability are generic properties of communication networks.

Journal Article↗