Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Neural network”

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 1,423 records · Page 79Linked to original sources
PubMed2004

On-line estimation of biomass concentration using a neural network and information about metabolic state.

This paper deals with the design of a neural network-based biomass concentration estimation system. This system is enhanced by the incorporation of information about the actual metabolism of the microorganism cultivated, which is taken from an on-line knowledge-based system. Two different design approaches have been investigated using the fed-batch cultivation of baker's yeast as the model process. In the first, metabolic state (MS) data were passed as additional input to the neural network; in the second, these data were used to select a neural network suitable for the specific MS. Two neural network types--feed-forward (Levenberg-Marquardt) and cascade correlation--were applied to this system and tested, and the performances of these neural networks were compared.

Algorithms↗
PubMed2004

New training strategies for constructive neural networks with application to regression problems.

Regression problem is an important application area for neural networks (NNs). Among a large number of existing NN architectures, the feedforward NN (FNN) paradigm is one of the most widely used structures. Although one-hidden-layer feedforward neural networks (OHL-FNNs) have simple structures, they possess interesting representational and learning capabilities. In this paper, we are interested particularly in incremental constructive training of OHL-FNNs. In the proposed incremental constructive training schemes for an OHL-FNN, input-side training and output-side training may be separated in order to reduce the training time. A new technique is proposed to scale the error signal during the constructive learning process to improve the input-side training efficiency and to obtain better generalization performance. Two pruning methods for removing the input-side redundant connections have also been applied. Numerical simulations demonstrate the potential and advantages of the proposed strategies when compared to other existing techniques in the literature.

Algorithms↗
PubMed2002

Optimization of the kernel functions in a probabilistic neural network analyzing the local pattern distribution.

This article proposes a procedure for the automatic determination of the elements of the covariance matrix of the gaussian kernel function of probabilistic neural networks. Two matrices, a rotation matrix and a matrix of variances, can be calculated by analyzing the local environment of each training pattern. The combination of them will form the covariance matrix of each training pattern. This automation has two advantages: First, it will free the neural network designer from indicating the complete covariance matrix, and second, it will result in a network with better generalization ability than the original model. A variation of the famous two-spiral problem and real-world examples from the UCI Machine Learning Repository will show a classification rate not only better than the original probabilistic neural network but also that this model can outperform other well-known classification techniques.

Neural Networks, Computer↗
PubMed2001

[Application of artificial neural network to simultaneous spectrophotometric determination of Cu, Co and Ni].

By means of artificial neural network and Levenberg-Marquardt back-propagation train algorithm, the three-component metal coordinate compounds of PAR-Cu, Co, Ni were determined simultaneously, in which the spectra overlapped. In 452-552 nm, the absorbance(A) at 14 wavelength were taken as character of artificial neural network, and samples were arranged by method of orthogonal design. The mean recovery of Cu, Co, Ni were 99.96%, 99.99% and 99.97% respectively. The RSD of the results were 0.1%, 0.2% and 0.1% respectively. The results were better than those of other networks in training speed and the accuracy. In conclusion, the new network spectrophotometry is a good choice for resolving multicomponent.

Cobalt↗
PubMed1999

Prediction of pharmacokinetic parameters and the assessment of their variability in bioequivalence studies by artificial neural networks.

PURPOSE: The methodology of predicting the pharmacokinetic parameters (AUC, cmax, tmax) and the assessment of their variability in bioequivalence studies has been developed with the use of artificial neural networks. METHODS: The data sets included results of 3 distinct bioequivalence studies of oral verapamil products, involving a total of 98 subjects and 312 drug applications. The modeling process involved building feedforward/backpropagation neural networks. Models for pharmacokinetic parameter prediction were also used for the assessment of their variability and for detecting the most influential variables for selected pharmacokinetic parameters. Variables of input neurons based on logistic parameters of the bioequivalence study, clinical-biochemical parameters, and the physical examination of individuals. RESULTS: The average absolute prediction errors of the neural networks for AUC, cmax, and tmax prediction were: 30.54%, 39.56% and 30.74%, respectively. A sensitivity analysis demonstrated that for verapamil the three most influential variables assigned to input neurons were: total protein concentration, aspartate aminotransferase (AST) levels, and heart-rate for AUC, AST levels, total proteins and alanine aminotransferase (ALT) levels, for cmax, and the presence of food, blood pressure, and body-frame for tmax. CONCLUSIONS: The developed methodology could supply inclusion or exclusion criteria for subjects to be included in bioequivalence studies.

Administration, Oral↗
PubMed2000

Neural network modeling for estimation of partition coefficient based on atom-type electrotopological state indices

A method for predicting log P values for a diverse set of 1870 organic molecules has been developed based on atom-type electrotopological-state (E-state) indices and neural network modeling. An extended set of E-state indices, which included specific indices with a more detailed description of amino, carbonyl, and hydroxy groups, was used in the current study. For the training set of 1754 molecules the squared correlation coefficient and root-mean-squared error were r2 = 0.90 and RMS(LOO) = 0.46, respectively. Structural parameters which included molecular weight and 38 atom-type E-state indices were used as the inputs in 39-5-1 artificial neural networks. The results from multilinear regression analysis were r2 = 0.87 and RMS(LOO) = 0.55, respectively. For a test set of 35 nucleosides, 12 nucleoside bases, 19 drug compounds, and 50 general organic compounds (n = 116) not included in the training set, a predictive r2 = 0.94 and RMS = 0.41 were calculated by artificial neural networks. The results for the same set by multilinear regression were r2 = 0.86 and RMS = 0.72. The improved prediction ability of artificial neural networks can be attributed to the nonlinear properties of this method that allowed the detection of high-order relationships between E-state indices and the n-octanol/water partition coefficient. The present approach was found to be an accurate and fast method that can be used for the reliable estimation of log P values for even the most complex structures.

Journal Article↗
PubMed2006

Implementation of neural networks for classification of moss and lichen samples on the basis of gamma-ray spectrometric analysis.

Mosses and lichens have an important role in biomonitoring. The objective of this study is to develop a neural network model to classify these plants according to geographical origin. A three-layer feed-forward neural network was used. The activities of radionuclides ((226)Ra, (238)U, (235)U, (40)K, (232)Th, (134)Cs, (137)Cs and (7)Be) detected in plant samples by gamma-ray spectrometry were used as inputs for neural network. Five different training algorithms with different number of samples in training sets were tested and compared, in order to find the one with the minimum root mean square error. The best predictive power for the classification of plants from 12 regions was achieved using a network with 5 hidden layer nodes and 3,000 training epochs, using the online back-propagation randomized training algorithm. Implementation of this model to experimental data resulted in satisfactory classification of moss and lichen samples in terms of their geographical origin. The average classification rate obtained in this study was (90.7 +/- 4.8)%.

Bryophyta↗
PubMedDate not supplied

[Artificial neural networks applied to study allergic conjunctivitis screening questionnaire].

PURPOSE: To evaluate sensibility and specificity of a screening questionnaire with multivariable analysis, compare them and elaborate an artificial neural network for future screenings. METHODS: Observational, transversal study performed at UNIFESP, with 48 patients with allergic conjunctivitis and 54 children without the disease. Their age ranged between 3 and 14 years and there was no restriction related to gender, systemic allergy or treatment. The questionnaire was applied and multivariable statistical analysis was performed. Finally, an artificial neural network was elaborated. RESULTS: Mean age was 8.4 years (7-13) and male gender was more frequent (60.7%). Mean score was 10.04 (0-18), and it was higher in the study group (p < 0.001). Allergic diagnosis was increased with the inclusion of the fifth question in 68.8%. Kappa coefficient was low (0.337; p = 0.071) and showed no agreement between diagnosis made by the questionnaire and clinical examination. Only the question number five had good sensitivity (85.4%) and specificity (85.1%). The cutoff point to separate allergic patients was 10 (sensitivity = 77.08% and specificity = 79.63%). The artificial neural network predicted allergic diagnosis in 100% using 7 of the 15 existent items. CONCLUSIONS: An efficient model was developed using seven questions, in a manner that its application might be easy to large populations.

Adolescent↗
PubMed2002

Complex dynamics and the structure of small neural networks.

The discrete-time dynamics of small neural networks is studied empirically, with emphasis laid on non-trivial bifurcation scenarios. For particular two- and three-neuron networks interesting dynamical properties like periodic, quasi-periodic and chaotic attractors are observed, many of them co-existing for one and the same set of parameters. An appropriate equivalence class of networks is defined, describing them as parametrized dynamical systems with identical dynamical capacities. Combined symmetries in phase space and parameter space are shown to generate different representatives of such a class. Moreover, conditions on the connectivity structure are suggested, which guarantee the existence of complex dynamics for a considered equivalence class of network configurations.

Algorithms↗
PubMed2001

Conversion of laparoscopic cholecystectomy to open cholecystectomy in acute cholecystitis: artificial neural networks improve the prediction of conversion.

Laparoscopic cholecystectomy is now also performed for acute cholecystitis. In the presence of inflammatory conditions, technical difficulties leading to conversion to open cholecystectomy may occur and overshadow the advantages of the laparoscopic approach. Factors associated with these undue events combined with techniques capable of learning from them may help in determining when to completely avoid the laparoscopic procedure. In this study we determined predictors of conversion in acute cholecystitis and tested their predictive ability by means of statistical multivariate analysis and artificial neural networks. Between January 1994 and February 1997, 225 patients underwent laparoscopic cholecystectomy for acute cholecystitis. Preoperative and operative data were prospectively collected on standardized forms. The first 180 laparoscopically approached cases entered the training set, which was learned by both the statistical and the artificial neural networks methods. Conversion was first studied in relation to a set of preoperative data. Prediction models were then fitted by both of these methods. The last 45 operated cases, which remained unknown to the learning systems, served for testing the fitted models. The forward stepwise logistic regression technique, the forward stepwise linear discriminant analysis, and the artificial neural networks method enabled positive prediction of conversion in 0%, 27%, and 100% of the cases, and a negative prediction in 80%, 85.5%, and 97% respectively, in the training set. A positive prediction of conversion in 0%, 25%, and 67% of the cases, and a negative prediction in 82%, 88%, and 94%, respectively, in the untrained, validation set of patients. An artificial neural networks based model provides a practical tool for the prediction of successful laparoscopic cholecystectomies and their conversion. The high degree of certainty of prediction in untrained cases reveals its potential, and justifies, under appropriate conditions, the complete avoidance of laparoscopy and turning directly to open cholecystectomy.

Acute Disease↗
PubMed2003

Enhanced prediction of lopinavir resistance from genotype by use of artificial neural networks.

Our objective was to accurately predict, from complex mutation patterns, human immunodeficiency virus type 1 resistance to the protease inhibitor lopinavir, by use of artificial intelligence. Two neural network models were constructed: 1 based on changes at 11 positions in the protease that were previously recognized as being significant for lopinavir resistance and another based on a newly derived set of 28 mutations that were identified by performing category prevalence analysis. Both models were trained, validated, and tested with 1322 clinical samples. A procedure of determining the optimal neural network parameters was proposed to speed up the training processes. The results suggested that the 28-mutation set was a more accurate predictor of lopinavir susceptibility (correlation coefficient, R2=0.88). We identified potentially significant new mutations associated with lopinavir resistance and demonstrated the utility of neural network models in predicting phenotypic susceptibility from complex genotypes.

Drug Resistance, Viral↗
PubMed2003

Application of neural networks in active noise reduction systems.

Active noise reduction systems based on a control unit in the form of a finite impulse response filter assume the linearity of every single component. Neural networks, which have so far been seldom used in this field, are a kind of a filter with the ability to project nonlinear characteristics of an active noise reduction system. This paper presents some simulation research studies of active noise reduction systems based on neural networks. Also presented are results of the operation of systems with different levels of complexity as well as the influence of different parameters of a neural network and of the system itself on those results.

Algorithms↗
PubMed1995

Two types of reorganization in multifunctional neural networks of Helix pomatia L.

The visceral functions of the snail Helix pomatia: Were shown to be regulated by overlapping and variable neural networks consisting of multifunctional neurons distributed in various parts of the CNS. The results showed that: 1. reorganization of the neural networks is accompanied by alterations in the firing mode of the neurons; 2. most identified neurons change their firing pattern from a tonic to a phasic type depending on the combination of signal molecules present or on the activation of their synaptic inputs; 3.the appearance of a slow membrane oscillation or burst firing was shown to make the neurons insensitive to input activation originating from visceral organs; 4. inputs from various visceral organs converge on a large number of neurons in the temporarily formed neural networks. In the vicinity of the multifunctional neurons a whole group of neurons can store various forms of transitory patterns; 5. both voltage (IK and ICa) and ligand (Ach and GABA) activated currents can be targets of signal molecules in altering the firing mode of the neurons. This will result in the reconfiguration of circuits; 6. because of to the large number of neurons involved in the regulation of visceral organs, the overlapping and reorganizing circuit can be interpreted more efficiently as a distributed network also including elements of dedicated regulatory circuits.

Animals↗
PubMed2005

The loading problem for recursive neural networks.

The present work deals with one of the major and not yet completely understood topics of supervised connectionist models. Namely, it investigates the relationships between the difficulty of a given learning task and the chosen neural network architecture. These relationships have been investigated and nicely established for some interesting problems in the case of neural networks used for processing vectors and sequences, but only a few studies have dealt with loading problems involving graphical inputs. In this paper, we present sufficient conditions which guarantee the absence of local minima of the error function in the case of learning directed acyclic graphs with recursive neural networks. We introduce topological indices which can be directly calculated from the given training set and that allows us to design the neural architecture with local minima free error function. In particular, we conceive a reduction algorithm that involves both the information attached to the nodes and the topology, which enlarges significantly the class of the problems with unimodal error function previously proposed in the literature.

Algorithms↗
PubMed2002

Applications of artificial neural network in AIDS research and therapy.

In recent years considerable effort has been devoted to applying pattern recognition techniques to the complex task of data analysis in drug research. Artificial neural networks (ANN) methodology is a modeling method with great ability to adapt to a new situation, or control an unknown system, using data acquired in previous experiments. In this paper, a brief history of ANN and the basic concepts behind the computing, the mathematical and algorithmic formulation of each of the techniques, and their developmental background is presented. Based on the abilities of ANNs in pattern recognition and estimation of system outputs from the known inputs, the neural network can be considered as a tool for molecular data analysis and interpretation. Analysis by neural networks improves the classification accuracy, data quantification and reduces the number of analogues necessary for correct classification of biologically active compounds. Conformational analysis and quantifying the components in mixtures using NMR spectra, aqueous solubility prediction and structure-activity correlation are among the reported applications of ANN as a new modeling method. Ranging from drug design and discovery to structure and dosage form design, the potential pharmaceutical applications of the ANN methodology are significant. In the areas of clinical monitoring, utilization of molecular simulation and design of bioactive structures, ANN would make the study of the status of the health and disease possible and brings their predicted chemotherapeutic response closer to reality.

Acquired Immunodeficiency Syndrome↗
PubMed2003

Risk factors for coronary artery disease and the use of neural networks to predict the presence or absence of high blood pressure.

BACKGROUND: The Framingham Heart Study was initiated in 1948 as a long-term longitudinal study to identify risk factors associated with cardiovascular disease (CVD). Over the years the scope of the study has expanded to include offspring and other family members of the original cohort, marker data useful for gene mapping and information on other diseases. As a result, it is a rich resource for many areas of research going beyond the original goals. As part of the Genetic Analysis Workshop 13, we used data from the study to evaluate the ability of neural networks to use CVD risk factors as training data for predictions of normal and high blood pressure. RESULTS: Applying two different strategies to the coding of CVD risk data as risk factors (one longitudinal and one independent of time), we found that neural networks could not be trained to clearly separate individuals into normal and high blood pressure groups. When training was successful, validation was not, suggesting over-fitting of the model. When the number of parameters was reduced, training was not as good. An analysis of the input data showed that the neural networks were, in fact, finding consistent patterns, but that these patterns were not correlated with the presence or absence of high blood pressure. CONCLUSION: Neural network analysis, applied to risk factors for CVD in the Framingham data, did not lead to a clear classification of individuals into groups with normal and high blood pressure. Thus, although high blood pressure may itself be a risk factor for CVD, it does not appear to be clearly predictable using observations from a set of other CVD risk factors.

Adult↗
PubMed1994

Classification by neural networks of evoked potentials. A first case study.

This paper describes the application of artificial neural networks for the analysis of data of the evoked potential type. A discussion of different preprocessing schemes stresses the importance of a suitable method which supports the artificial neural networks in their classification task. This preprocessing differs completely from well-known data reduction methods used for non-neural classifiers. The examples shown here show the extraordinarily good tolerance against noise, yielding very good classification results for only a small number of observations.

Alcoholism↗
PubMed1995

Neural network assisted cardiac auscultation.

Traditional cardiac auscultation involves a great deal of interpretive skill. Neural networks were trained as phonocardiographic classifiers to determine their viability in this rôle. All networks had three layers and were trained by backpropagation using only the heart sound amplitude envelope as input. The main aspect of the study was to determine what topologies, gain and momentum factors lead to efficient training for this application. Neural networks which are trained with heart sound classes of greater similarity were found to be less likely to converge to a solution. A prototype normal/abnormal classifier was also developed which provided excellent classification accuracy despite the sparse nature of the training data. Future directions for the development of a full-scale computer-assisted phonocardiographic classifier are also considered.

Cardiology↗