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 1,027 records · Page 57Linked to original sources
PubMed1994

Sulfate detection in glycoprotein-derived oligosaccharides by artificial neural network analysis of Fourier-transform infrared spectra.

We report the use of an artificial neural network to analyze the fingerprint region of Fourier-transform infrared (ir) spectra of oligosaccharides for the presence of sulfate groups. This assay can rapidly and nondestructively detect the presence of sulfate in as little as 1 nmol (approximately 2 micrograms) of a glycoprotein-derived monosulfated decasaccharide. The neural network was trained to recognize the presence of sulfate groups by presenting it with 45 ir spectra of sulfated and nonsulfated mono- and oligosaccharides. No prior knowledge of the characteristic ir spectral features of a sulfate group was needed as input. The training process required between 3 and 10 h, while analysis of a spectrum with the trained neural network requires only 0.1 s.

Carbohydrate Sequence↗
PubMed2006

Generalized RLS approach to the training of neural networks.

Recursive least square (RLS) is an efficient approach to neural network training. However, in the classical RLS algorithm, there is no explicit decay in the energy function. This will lead to an unsatisfactory generalization ability for the trained networks. In this paper, we propose a generalized RLS (GRLS) model which includes a general decay term in the energy function for the training of feedforward neural networks. In particular, four different weight decay functions, namely, the quadratic weight decay, the constant weight decay and the newly proposed multimodal and quartic weight decay are discussed. By using the GRLS approach, not only the generalization ability of the trained networks is significantly improved but more unnecessary weights are pruned to obtain a compact network. Furthermore, the computational complexity of the GRLS remains the same as that of the standard RLS algorithm. The advantages and tradeoffs of using different decay functions are analyzed and then demonstrated with examples. Simulation results show that our approach is able to meet the design goals: improving the generalization ability of the trained network while getting a compact network.

Algorithms↗
PubMed2000

Prediction on lengths of stay in the postanesthesia care unit following general anesthesia: preliminary study of the neural network and logistic regression modelling.

The length of stay in the postanesthesia care unit (PACU) following general anesthesia in adults is an important issue. A model, which can predict the results of PACU stays, could improve the utilization of PACU and operating room resources through a more efficient arrangement. The purpose of study was to compare the performance of neural network to logistic regression analysis using clinical sets of data from adult patients undergoing general anesthesia. An artificial neural network was trained with 409 clinical sets using backward error propagation and validated through independent testing of 183 records. Twenty-two inputs were used to find determinants and to predict categorical values. Logistic regression analysis was performed to provide a comparison. The neural network correctly predicted in 81.4% of situations and identified discriminating variables (intubated state, sex, neuromuscular blocker and intraoperative use of opioid), whereas the figure was 65.0% in logistic regression analysis. We concluded that the neural network could provide a useful predictive model for the optimization of limited resources. The neural network is a new alternative classifying method for developing a predictive paradigm, and it has a higher classifying performance compared to the logistic regression model.

Adult↗
PubMed2001

Sensor fusion by neural networks using spatially represented information.

A neural network model based on a lateral-inhibition-type feedback layer is analyzed with regard to its capabilities to fuse signals from two different sensors reporting the same event ("multisensory convergence"). The model consists of two processing stages. The input stage holds spatial representations of the sensor signals and transmits them to the second stage where they are fused. If the input signals differ, the model exhibits two different processing modes: with small differences it produces a weighted average of the input signals, whereas with large differences it enters a decision mode where one of the two signals is suppressed. The dynamics of the network can be described by a series of two first-order low-pass filters, whose bandwidth depends nonlinearly on the level of concordance of the input signals. The network reduces sensor noise by means of both its averaging and filtering properties. Hence noise suppression, too, depends on the level of concordance of the inputs. When the network's neurons have internal noise, sensor noise suppression is reduced but still effective as long as the input signals do not differ strongly. The possibility of extending the scheme to three and more inputs is discussed.

Artifacts↗
PubMedDate not supplied

The influence of lifestyle on cardiovascular risk factors. Analysis using a neural network.

The cardiovascular pathologies are the most common causes of death in the elderly patient. To single out the main risk factors in order to effectively prevent the onset of the disease, the authors experimented a special computerized tool, the neural network, that works out a mathematical relation that can obtain certain data (defined as output) as a function of other data (defined as input). Data were processed from a sample of 276 subjects of both sexes aged 26-69 years old. The output data were: high/low cholesterolemia, HDL cholesterol, triglyceridemia with respect to an established cut-off; the input data were: sex, age, build, weight, married/single, number of children, number of cigarettes smoked/day, amount of wine and number of cups of coffee. We conclude that: (i) a relationship exists, deduced from a neural network, between a set of input variables and a dichotomous output variable; (ii) this relationship can be expressed as a mathematical function; (iii) a neural network, having learned the data on a sufficiently large population, can provide valid predictive data for a single individual with a high probability (up to 93.33%) that the response it gives is correct. In this study, such a result is found for two of the three cardiovascular risk indicators considered (cholesterol and triglycerides); (iv) the repetition of the neural network analysis of the cases in question after a "pruning" operation provided a somewhat less good performance; (v) a statistical analysis conducted on those same cases has confirmed the existence of a strong relationship between the input and the output variables. Therefore the neural network is a valid instrument for providing predictive in a single subject on cardiovascular pathology risks.

Adult↗
PubMed1998

[Library search of UV spectra of organic environmental pollutants by artificial neural network].

In this paper, the artificial neural network (ANN) was applied to the library search of UV spectra of organic environmental pollutants. The optimization of network parameters was discussed and the effects of noise and impurity were investigated in detail. The use of derivative spectra for ANN library search was proposed to enhance the resolution of the UV spectra. This method could speed up the convergence of the network and could enhance the tolerance of impurity level, but was subject to noise comparing with the ANN using conventional UV spectra. Results show that the ANN is superior to the correlation coefficient method in resistance to noise and impurity.

Databases, Factual↗
PubMed1996

Using neural networks to model conditional multivariate densities.

Neural network outputs are interpreted as parameters of statistical distributions. This allows us to fit conditional distributions in which the parameters depend on the inputs to the network. We exploit this in modeling multivariate data, including the univariate case, in which there may be input-dependent (e.g., time-dependent) correlations between output components. This provides a novel way of modeling conditional correlation that extends existing techniques for determining input-dependent (local) error bars.

Economics↗
PubMed1997

Using an artificial neural network to define the planning target volume in radiotherapy.

A neural network for predicting the planning target volume in radiotherapy from the shape of the detected tumor is designed and tested in this research project. The proposed neural network is able to generalize expert medical knowledge and predict the planning target volume from a three-dimensional image of the detected tumor. Initial results for simple shaped brain tumors are presented in this paper.

Artificial Intelligence↗
PubMed1995

The assessment of LH surge for predicting ovulation time using clinical, hormonal, and ultrasonic indices in infertile women with an ensemble of neural networks.

An ensemble of independently trained neural networks (NN) is proposed for the assessment of luteinizing hormone (LH) surge for predicting ovulation time in infertile but ovulating women. The proposed ensemble involves a number of parallel NN modules. Each pair of the NNs learn specific data that are previously collected for monitoring timing function of LH estradiol (pg ml-1), and follicle diameter (mm) are used to train NN pairs to approximate the function of the LH values. A reasonable and accurate estimation places ovulation approximately 10-12 h after the LH peak. The double-valued (bi-phasic) regions of training data are separated into two single-valued (bi-phasic) regions of training data are separated into two single-valued parts (not exactly preovulatory, postovulatory division) that can be learned by each module of the NN pair. During testing, after the initial decision to have single-valued sides, the assessment is obtained by a linear opinion pool (consensus rule) using the decisions of NNs on the corresponding side without waiting. The network ensemble has various desirable properties: high assessment accuracy of a double-valued multisource data, minimized learning and recall times, and a parallel structure. The ovulation time can be predicted through the assessment of LH peak with a better precision and fewer number of tests.

Diagnosis, Computer-Assisted↗
PubMed1996

The interchangeability of learning rate and gain in backpropagation neural networks.

The backpropagation algorithm is widely used for training multilayer neural networks. In this publication the gain of its activation function(s) is investigated. In specific, it is proven that changing the gain of the activation function is equivalent to changing the learning rate and the weights. This simplifies the backpropagation learning rule by eliminating one of its parameters. The theorem can be extended to hold for some well-known variations on the backpropagation algorithm, such as using a momentum term, flat spot elimination, or adaptive gain. Furthermore, it is successfully applied to compensate for the nonstandard gain of optical sigmoids for optical neural networks.

Algorithms↗
PubMed1991

Designing a neural network simulator--the MENS modelling environment for network systems: II.

During recent years, neural network research has been extended to a large number of different fields, increasingly attracting the interest of workers from various disciplines. The computer simulations carried out with this research require an appropriate software environment. The computational similarities of many kinds of simulations allow the design of software components that are largely independent of the specific application. These considerations are reflected, for example, by the general layout of the MENS network simulator, as described in the accompanying first paper. This paper presents the design considerations for the simulator's different software components in more detail. In particular, design and implementation are discussed with respect to computational and memory efficiency. The discussion includes, for example, the representation of a network by the simulator's data structure, the file-driven configuration and initialization of a network, the simulator's stimulus and monitor system, and the simulator's control structures. In addition, the separation and interaction of application-specific and application-independent software components are addressed. Particular performance aspects comprise the implementation of synaptic delays, the dynamic deletion of synaptic links in network learning, and the preprocessing of stimulus films. In addition, some general aspects of simulator performance and testing are considered. The material presented in this paper concerns both the development of new simulation software and the efficient use of existing programs. Therefore, both the general user as well as the software designer may hopefully benefit from this presentation.

Animals↗
PubMedDate not supplied

Use of a probabilistic neural network to estimate the risk of mortality after cardiac surgery.

OBJECTIVE: To develop a probabilistic neural network (PNN) to estimate mortality risk following cardiac surgery. DESIGN AND SETTING: The PNN model was created using an institutional database obtained as part of routine quality assurance activity. Patient records (from 1991 to 1993) were randomly divided into training (n = 732) and validation (n = 380) sets. The model uses seven variables, each obtainable during routine clinical patient care. After completion of the initial validation phase, newer data (1994) became available and were used as an independent source of validation (n = 365). PATIENTS: 1,477 consecutive cardiac surgery patients operated on in a teaching hospital during a four-year period (1991-94). RESULTS: The overall accuracy of the neural network was 91.5% in the training set; it was 92.3% in the validation set. The model was well calibrated (p = 0.21 for the Hosmer-Lemeshow goodness-of-fit test) and discriminated well (areas under the ROC curves were 0.72 and 0.81 for the training and validation sets). The trained network also performed well on the 1994 data (ROC = 0.74, p = 0.19 for the Hosmer-Lemeshow test), albeit with a slight decrement in overall accuracy (88.2%). CONCLUSION: A neural network may be implemented to estimate mortality risk following cardiac surgery. Implementation is relatively rapid, and it is an alternative to standard statistical approaches.

Adult↗
PubMed2000

Viscosity prediction of lipophilic semisolid emulsion systems by neural network modelling.

Previously published data (Gasperlin et al., 1998) on viscoelastic behaviour of lipophilic semisolid emulsion systems and the prediction of their physical stability by neural network modelling are analysed in further detail. Most attention has been paid to viscosity, which with storage (G') and loss modulus (G"), is one of the most important rheological parameters influenced by structure. Complex dynamic viscosity (eta*) was measured by oscillatory rheometry. The viscosity dependence of the lipophilic semisolid emulsions on the ratio of the particular components was defined by the neural network (error back-propagation algorithm), linear and incomplete polynomial models of higher orders. Polynomial models were used to complement the neural network model and to determine the relationship between variables. Since the viscosity was expressed in the whole measured frequency range, modelling was more complex and indirect modelling was introduced. The determined models were tested and the results confirm their usefulness for the explanation and prediction of the rheological characteristics of emulsion systems. The trained and tested neural network model proved to be a highly effective and applicable tool for predicting the viscosity of a lipophilic semisolid emulsion system of given composition.

Emulsions↗
PubMed1993

Digital neural networks for high-speed divisions and root extractions.

The instruction set of digital integer-valued VLSIs implementing neural networks is typically restricted to weighted sums, threshold activation functions and weight assignments. However, neural networks also need divisions (for data normalizations) and root extractions (for distance or vector norm computations). This paper presents neural networks which perform the latter operations by using only the above basic instructions. These networks thus provide "macros" to the overall network which calls them. This yields a homogeneous environment. Moreover, a "virtual base" is introduced to exploit the available parallelism. It is freely chosen so as to achieve the desired trade-off between the speed and complexity of the proposed networks (with a minimum of only two neurons). This high-speed capability also makes these structures attractive independently from neural applications.

Algorithms↗
PubMed1992

Neural network synthesis of spin echo multiecho sequences.

Spin echo multiecho sequences are not frequently used in clinical practice, because they allow the observation of one single slice, imaged at different echo times, for each acquisition. To limit examination time, multislice sequences that include only images derived from one or two echoes are usually acquired. Nevertheless, the strong T2 dependence of multiecho sequences can be used effectively to enhance the contrast between tissues with different T2 and to gather useful diagnostic information. Artificial neural networks can offer new interesting facilities to the radiologist. In fact, the learning capabilities of neural networks allow them to extract the prototypical behavior of a system from a set of examples. After learning, artificial neural networks can emulate the system behavior even in the presence of new inputs, as far as these are not too different from those included in the training set. A conveniently trained neural network can synthesize a multiecho sequence for each slice of a multislice sequence, requiring only two images for each slice to achieve reliable results. When compared with a true multiecho sequence, the images generated by the network preserve the contrast characteristics of the original ones and have a better signal-to-noise (SNR) ratio. In this paper we report the results achieved by using a neural network to reconstruct synthetic spin echo multiecho images of the brain.

Image Processing, Computer-Assisted↗
PubMedDate not supplied

'Neural network' algorithm to predict severity in epidermolysis bullosa simplex.

BACKGROUND AND AIMS: There are various genotypic variations known for the four phenotypic presentations of epidermolysis bullosa simplex (EBS). A neural network algorithm may be used to find the relationship between the various factors responsible for a particular phenotypic presentation. We assessed the value of neural network to predict the prognosis of epidermolysis bullosa simplex. METHODS: Cases of EBS in which sequencing had been performed to identify the position and type of mutation were collected by literature search and the resulting data was analyzed using neural network algorithm. RESULTS: The statistical prediction had an accuracy rate of 78%. CONCLUSION: Neural networks can identify hidden patterns in a huge database without the intervention of a skilled statistician. It has the potential to change the way we analyze clinical and experimental data at present.

Algorithms↗
PubMedDate not supplied

Diagnosis of sensorineural hearing loss with neural networks versus logistic regression modeling of distortion product otoacoustic emissions.

We investigated whether modeling with artificial neural networks or logistic regression of distortion product otoacoustic emissions (DPOAE), across diverse frequencies, may achieve an accurate diagnosis of sensorineural hearing loss (SNHL) of cochlear origin. 256 ears (90 with SNHL and 166 with normal hearing) were evaluated with pure-tone audiometry, impedance audiometry, speech audiometry and DPOAE. Ears were split into training (n = 176) and validation (n = 80) sets. Input variables included gender, age, examination time, DPOAE intensity at F(2) frequencies 593, 937, 1906, 3812 and 6031 Hz, and respective values corrected for noise levels. In the validation data set, an average network had an area under the receiver operating characteristic curve (AUC) of 0.86 (accuracy 84%). Logistic regressions including all these variables or those selected by backward elimination had AUC values of 0.91 and 0.92, respectively (accuracy 85% both). Eleven of 12 trained networks had better specificity than the backward elimination logistic regression, and the backward elimination logistic regression had a better sensitivity than 11 of the 12 networks. Both modeling approaches correctly identified all ears with sudden hearing loss, congenital hearing loss, head trauma, nuclear jaundice and ototoxicity, and 2-3 of 5 ears with acoustic trauma, but missed 1-3 of 3 ears with Ménière's disease and 4-6 of 8 ears with abnormal pure-tone thresholds on audiometry which had no accompanying findings. For SNHL exceeding 45 dB HL on a pure-tone threshold, sensitivity was 83% (15/18) by neural networks and 84 or 94% (16/18 or 17/18) by logistic regression. Both neural-network-based analysis and logistic regression modeling of the DPOAE pattern across a range of frequencies offer promising approaches for the objective diagnosis of moderate and severe SNHL.

Adult↗
PubMed1995

[Learning and prediction of a chaotic pattern related to the time-course of clinical laboratory data by artificial neural network].

We investigated whether an artificial neural network (ANN) could learn a chaotic pattern related to the time-course of some clinical laboratory data and predict its future pattern. Then, the patterns of serum ALT level over time in patients with chronic hepatitis were successively learned by ANN. Future patterns were also successively predicted. This procedure may be used to pick up an unexpected event in time-course of personal data, study the pathological background of disease, or application of prophylactic treatment.

Alanine Transaminase↗