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 739 records · Page 41Linked to original sources
PubMed2006

An approach to estimate EEG power spectrum as an index of heat stress using backpropagation artificial neural network.

A method has been presented for an effective application of backpropagation artificial neural network (ANN) in establishment of electro-encephalogram (EEG) power spectra as an index of stress in hot environment. The power spectrum data for slow wave sleep (SWS), rapid eye movement (REM) sleep and awake (AWA) states in three groups of rats (acute heat stress, chronic heat stress and the normal) were tested by an ANN, containing 60 nodes in input layer, weighted from power spectrum data from 0 to 30 Hz, 18 nodes in hidden layer and an output node. The target output values for this network were determined with another five-layered neural network (with the structure of 3-12-1-12-3). The input and output of this network was assigned with the three well-established heat stress indices (body temperature, body weight and plasma corticosterone). The most important feature for acute stress, chronic stress and normal conditions were extracted from the third layer single neuron and used for the target value for the three-layered neural network. The ANN was found effective in recognising the EEG power spectra with an average of 96.67% for acute heat stress, 97.17% for chronic heat stress and 98.5% for normal subjects.

Algorithms↗
PubMed2001

Cellular neural network analysis for two-dimensional bioheat transfer equation.

The cellular neural network (CNN) method is applied to solve the Pennes bioheat transfer equation, and its feasibility is demonstrated. Numerical solutions were obtained for a cellular neural network for a two-dimensional steady-state temperature field obtained from focused and unfocused ultrasound heat sources. Transient-state temperature fields were also studied and compared with experimental results obtained elsewhere. The cellular neural networks' key features of asynchronous parallel processing, continuous-time dynamics and local interaction enable real-time temperature field estimation for clinical hyperthermia.

Energy Transfer↗
PubMed2006

Toward the training of feed-forward neural networks with the D-optimum input sequence.

The problem under consideration is to obtain a measurement schedule for training neural networks. This task is perceived as an experimental design in a given design space that is obtained in such a way as to minimize the difference between the neural network and the system being considered. This difference can be expressed in many different ways and one of them, namely, the D-optimality criterion is used in this paper. In particular, the paper presents a unified and comprehensive treatment of this problem by discussing the existing and previously unpublished properties of the optimum experimental design (OED) for neural networks. The consequences of the above properties are discussed as well. A hybrid algorithm that can be used for both the training and data development of neural networks is another important contribution of this paper. A careful analysis of the algorithm is presented and its comprehensive convergence analysis with the help of the Lyapunov method are given. The paper contains a number of numerical examples that justify the application of the OED theory for neural networks. Moreover, an industrial application example is given that deals with the valve actuator.

Algorithms↗
PubMed1997

Linear and nonlinear ARMA model parameter estimation using an artificial neural network.

This paper addresses parametric system identification of linear and nonlinear dynamic systems by analysis of the input and output signals. Specifically, we investigate the relationship between estimation of the system using a feedforward neural network model and estimation of the system by use of linear and nonlinear autoregressive moving-average (ARMA) models. By utilizing a neural network model incorporating a polynomial activation function, we show the equivalence of the artificial neural network to the linear and nonlinear ARMA models. We compare the parameterization of the estimated system using the neural network and ARMA approaches by utilizing data generated by means of computer simulations. Specifically, we show that the parameters of a simulated ARMA system can be obtained from the neural network analysis of the simulated data or by conventional least squares ARMA analysis. The feasibility of applying neural networks with polynomial activation functions to the analysis of experimental data is explored by application to measurements of heart rate (HR) and instantaneous lung volume (ILV) fluctuations.

Computer Simulation↗
PubMed2002

Computerized detection of supporting forelimb lameness in the horse using an artificial neural network.

The purpose of this study was to investigate whether artificial neural networks could be used to determine equine lameness by computational means only. The integral parts of our approach were the combination of automated signal tracking of horses on a treadmill and the computational power of artificial neural networks (ANN). The motion of 175 horses trotting on a treadmill was recorded using the SELSPOT II system for motion analysis. Two cameras traced infrared (IR) markers on the head and on the left forehoof. The motion of the head was Fourier-transformed and further processed by a multilayer feedforward ANN, which was trained to distinguish healthy from pathological gaits and to quantify the lameness. The classification was correct in 78.6% of cases. In 12% of cases the network gave contradictory results, in 5.9% the network found no answers, and in 3.5% the answers were wrong. However after proper training, it is proposed that neural networks are potentially capable of making a non-human diagnosis of equine lameness.

Animals↗
PubMed1990

Artificial neural networks and their use in quantitative pathology.

A brief general introduction to artificial neural networks is presented, examining in detail the structure and operation of a prototype net developed for the solution of a simple pattern recognition problem in quantitative pathology. The process by which a neural network learns through example and gradually embodies its knowledge as a distributed representation is discussed, using this example. The application of neurocomputer technology to problems in quantitative pathology is explored, using real-world and illustrative examples. Included are examples of the use of artificial neural networks for pattern recognition, database analysis and machine vision. In the context of these examples, characteristics of neural nets, such as their ability to tolerate ambiguous, noisy and spurious data and spontaneously generalize from known examples to handle unfamiliar cases, are examined. Finally, the strengths and deficiencies of a connectionist approach are compared to those of traditional symbolic expert system methodology. It is concluded that artificial neural networks, used in conjunction with other nonalgorithmic artificial intelligence techniques and traditional algorithmic processing, may provide useful software engineering tools for the development of systems in quantitative pathology.

Artificial Intelligence↗
PubMed1996

Use of neural networks in brain SPECT to diagnose Alzheimer's disease.

UNLABELLED: The usefulness of artificial neural networks in the classification of 99mTc-HMPAO SPECT axial brain scans was investigated in a study group of Alzheimer's disease patients and age-matched normal subjects. METHODS: The cortical circumferential profiling (CCP) technique was used to extract information regarding patterns of cortical perfusion. Traditional analysis of the CCP data, taken from slices at the level of the basal ganglia, indicated significant perfusion deficits for Alzheimer's disease patients relative to normals, particularly in the left temporo-parietal and left posterior frontal areas of the cortex. The compressed profiles were then used to train a neural-network classifier, the performance of which was compared with that of a number of more traditional statistical (discriminant function) techniques and that of two expert viewers. RESULTS: The optimal classification performance of the neural network (ROC area = 0.91) was better than that of the alternative statistical techniques (max. ROC area = 0.85) and that of the expert viewers (max. ROC area = 0.79). CONCLUSION: The CCP produces perfusion profiles which are well suited to automated classification methods, particularly those employing neural networks. The technique has the potential for wide application.

Aged↗
PubMed2005

Effects of input data on the performance of a neural network in distinguishing normal and glaucomatous visual fields.

PURPOSE: To compare the performance of neural networks for perimetric glaucoma diagnosis when using different types of data inputs: numerical threshold sensitivities, Statpac Total Deviation and Pattern Deviation, and probability scores based on Total and Pattern Deviation probability maps (Carl Zeiss Meditec, Inc., Dublin, CA). METHODS: The results of SITA Standard visual field tests in 213 healthy subjects, 127 patients with glaucoma, 68 patients with concomitant glaucoma and cataract, and 41 patients with cataract only were included. The five different types of input data were entered into five identically designed artificial neural networks. Network thresholds were adjusted for each network. Receiver operating characteristic (ROC) curves were constructed to display the combinations of sensitivity and specificity. RESULTS: Input data in the form of Pattern Deviation probability scores gave the best results, with an area of 0.988 under the ROC curve, and were significantly better (P < 0.001) than threshold sensitivities and numerical Total Deviations and Total Deviation probability scores. The second best result was obtained with numerical Pattern Deviations with an area of 0.980. CONCLUSIONS: The choice of type of data input had important effects on the performance of the neural networks in glaucoma diagnosis. Refined input data, based on Pattern Deviations, resulted in higher sensitivity and specificity than did raw threshold values. Neural networks may have high potential in the production of useful clinical tools for the classification of visual field tests.

Adult↗
PubMed1997

Computational neural networks for predictive microbiology. II. Application to microbial growth.

The growth of a specific microorganism on a certain food is influenced by a number of environmental factors such as temperature, pH, and salt concentration. Methods that delineate the history of the growth of microorganisms are always subject to a considerable debate and scrutiny in the field of predictive microbiology. Regardless of its types, a growth model (e.g., modified Gompertz model) contains several parameters that vary depending on the microorganisms/food combination and the associated prevailing environmental conditions. The growth model parameters for a set of operating conditions are commonly determined from expressions developed via multiple linear regressions. In the present study, a substitute for the nonlinear regression-based equations is developed using computational neural networks. Computational neural networks are applied herein on experimental data pertaining to the anaerobic growth of Shigella flexneri. Results have indicated that predictions by neural networks offer better agreement with experimental data as compared to predictions obtained via corresponding regression equations.

Neural Networks, Computer↗
PubMed2002

Prediction of joint moments using a neural network model of muscle activations from EMG signals.

Because the relationship between electromyographic (EMG) signals and muscle activations remains unpredictable, a new way to determine muscle activations from EMG signals by using a neural network is proposed and realized. Using a neural network to predict the muscle activations from EMG signals avoids establishing a complex mathematical model to express the muscle activation dynamics. The feed-forward neural network model of muscle activations applied here is composed of four layers and uses an adjusted back-propagation training algorithm. In this study, the basic back-propagation algorithm was not applicable, because muscle activation could not be measured, and hence the error between predicted activation and the real activation was not available. Thus, an adjusted back-propagation algorithm was developed. Joint torque at the elbow was calculated from the EMG signals of ten flexor and extensor muscles, using the neural network result of estimated activation of the muscles. Once muscle activations were obtained, Hill-type models were used to estimate muscle force. A musculoskeletal geometry model was then used to obtain moment arms, from which joint moments were determined and compared with measured values. The results show that this neural network model can be used to represent the relationship between EMG signals and joint moments well.

Action Potentials↗
PubMed1996

A higher order Bayesian neural network with spiking units.

We treat a Bayesian confidence propagation neural network, primarily in a classifier context. The one-layer version of the network implements a naive Bayesian classifier, which requires the input attributes to be independent. This limitation is overcome by a higher order network. The higher order Bayesian neural network is evaluated on a real world task of diagnosing a telephone exchange computer. By introducing stochastic spiking units, and soft interval coding, it is also possible to handle uncertain as well as continuous valued inputs.

Action Potentials↗
PubMed2001

Introduction to artificial neural networks for physicians: taking the lid off the black box.

BACKGROUND: Over the past 5 years, a steady stream of publications has discussed the use of artificial neural networks (ANNs) for urologic and other medical applications. The pace of this research has increased recently, and deployed products based on this technology are now appearing. Before these tools can be widely accepted by clinicians and researchers, a deeper level of understanding of ANNs is necessary. This article attempts to lay some of the groundwork needed to facilitate this familiarity. METHODS: A short discussion of neural network history is included for background. This is followed by an in-depth discussion of how and why ANNs work. This discussion includes the relationship between ANNs and statistical regression. An investigation of issues associated with neural networks follows, applicable to both general and urologic-specific applications. RESULTS: Neural networks are computer models that have been studied extensively for over 50 years, with prostate cancer applications since 1994. From a biological viewpoint, ANNs are artificial analogues of data structures that exist in nervous systems. From a numeric viewpoint, ANNs are matrices of numbers whose values comprise knowledge that is distilled from historic databases. Many types of neural networks are analogous to well-known statistical methods. CONCLUSIONS: ANNs are complex numeric constructs, but no more complex than similar statistical methods. However, several issues associated with neural network derivation demand that developers apply rigorous engineering practices in their studies.

Humans↗
PubMed1994

A breathing circuit alarm system based on neural networks.

OBJECTIVE: The objectives of our study were (1) to implement intelligent respiratory alarms with a neural network; and (2) to increase alarm specificity and decrease false-alarm rates compared with current alarms. METHODS: We trained a neural network to recognize 13 faults in an anesthesia breathing circuit. The system extracted 30 breath-to-breath features from the airway CO2, flow, and pressure signals. We created training data for the network by introducing 13 faults repeatedly in 5 dogs (616 total faults). We used the data to train the neural network using the backward error propagation algorithm. RESULTS: In animals, the trained network reported the alarms correctly for 95.0% of the faults when tested during controlled ventilation, and for 86.9% of the faults during spontaneous breathing. When tested in the operating room, the system found and correctly reported 54 of 57 faults that occurred during 43.6 hr of use. The alarm system produced a total of 74 false alarms during 43.6 hr of monitoring. CONCLUSION: Neural networks may be useful in creating intelligent anesthesia alarm systems.

Anesthesiology↗
PubMed2005

[Identification of official rhubarb samples based on IR spectra and neural networks].

The Fourier transform infrared (IR) spectrometry and neural networks have been used to identification of official and un-official rhubarb samples in the present work. The IR spectra were compressed by using wavelet transform and then were normalized prior to network training. Spectra with 700 data points were compressed to 44 variables, therefore, the training process of neural networks were speed up. 52 rhubarb samples in which 25 official and 27 unofficial rhubarb samples are included have been used to network modeling. The effects of neuron number in hidden layer and momentum parameter on classification have been investigated. Results showed that about 98% rhubarb samples could be identified correctly when optimized parameters were used. This method can be useful for quality control in rhubarb-contained Chinese medicine production.

Drugs, Chinese Herbal↗
PubMed1998

Use of a neural network to predict stone growth after shock wave lithotripsy.

OBJECTIVES: To determine whether a neural network is superior to standard computational methods in predicting stone regrowth after shock wave lithotripsy (SWL) and to determine whether the presence of residual fragments, as an independent variable, increases risk. METHODS: We reviewed the records of 98 patients with renal or ureteral calculi treated by primary SWL at a single institution and followed up for at least 1 year; residual stone fragment growth or new stone occurrence was determined from abdominal radiographs. A neural network was programmed and trained to predict an increased stone volume over time utilizing input variables, including previous stone events, metabolic abnormality, directed medical therapy, infection, caliectasis, and residual fragments after SWL. Patient data were partitioned into a training set of 65 examples and a test set of 33. The neural network did not encounter the test set until training was complete. RESULTS: The average follow-up period was 3.5 years (range 1 to 10). Of 98 patients, 47 had residual stone fragments 3 months after SWL; of these 47, 8 had increased stone volume at last follow-up visit. Of 51 patients stone free after SWL, 4 had stone recurrence. Coexisting risk factors were incorporated into a neural computational model to determine which of the risk factors was individually predictive of stone growth. The classification accuracy of the neural model in the test set was 91%, with a sensitivity of 91%, a specificity of 92%, and a receiver operating characteristic curve area of 0.964, results significantly better than those yielded by linear and quadratic discriminant function analysis. CONCLUSIONS: A computational tool was developed to predict accurately the risk of future stone activity in patients treated by SWL. Use of the neural network demonstrates that none of the risk factors for stone growth, including the presence of residual fragments, is individually predictive of continuing stone formation.

Follow-Up Studies↗
PubMed2001

Application of classification tree and neural network algorithms to the identification of serological liver marker profiles for the diagnosis of hepatocellular carcinoma.

OBJECTIVE: Although many attempts have been made to identify tumour-specific alpha-fetoprotein (AFP) glycoforms or other serological markers for the diagnosis of hepatocellular carcinoma (HCC), none of the available markers has, so far, shown satisfactory sensitivity and specificity. Here we aimed to apply classification tree and neural network algorithms to interpret the levels of multiple serological liver markers to improve overall specificity and sensitivity, particularly with a view to discriminating between liver cirrhosis with and without HCC. METHODS: We developed classification trees and neural networks that identified serological liver marker profiles comprising AFP, alpha1-antitrypsin (A1AT), alpha2-macroglobulin (A2MG), thyroxine-binding globulin (TBG), transferrin and albumin as well as sex and age, which might permit the diagnosis of HCC. Data were collected from 65 HCC patients, 51 patients with liver cirrhosis alone (LC) and 51 normal healthy subjects. RESULTS: The generated classification trees and neural networks showed similar diagnostic values in differentiating HCC from LC. The classification trees identified AFP, A1AT and albumin as the most important classification parameters, whereas the neural networks identified A2MG, AFP, A1AT and albumin as the predominant factors. The classification logic of the classification trees indicated that more HCC cases could be identified among cases with slightly elevated AFP levels by using the serum levels of A1AT and albumin. The neural networks were also useful for the identification of the HCC cases when the AFP levels were below 500 ng/ml (p < 0.005). The neural networks could identify HCC cases with AFP levels within the normal range, but the classification trees could not. By combining the conventional AFP test and the neural networks, the overall diagnostic sensitivity for HCC was significantly increased from 60.0 to 73.8% (p < 0.05) while maintaining a high specificity (88.2%). The sensitivities for tumors of different sizes were similar. CONCLUSION: The neural network algorithm appeared to be more powerful than the classification tree algorithm in the identification of the distinctive serological liver marker profiles for the diagnosis of the HCC subgroup without significant elevation in serum AFP levels. By incorporating serological levels of other liver markers and including data from a large number of patients and control subjects, it should prove possible to develop a versatile neural network for early diagnosis of HCC.

Adult↗
PubMed1993

Optoelectronic implementation of neural networks.

The performance of electronic neural networks will eventually be limited by the capacity of their interconnections as operating speeds are increased. The use of optical connections to link electronic elements offers a way to overcome this problem. In particular, spatial optic techniques are especially suited to neural systems because of their parallel structure. Three experimental optoelectronic networks are described that employ holographic components and arrays of high-speed optical modulators. These of a network to 64 x 8 connections with optical fan-out. A hybrid networks have been used to demonstrate operation at 50 MHz, novel training algorithms and the expansion technique for integrating optoelectronic systems is also being developed.

Algorithms↗
PubMed2005

Comparison between logistic regression and neural networks to predict death in patients with suspected sepsis in the emergency room.

INTRODUCTION: Neural networks are new methodological tools based on nonlinear models. They appear to be better at prediction and classification in biological systems than do traditional strategies such as logistic regression. This paper provides a practical example that contrasts both approaches within the setting of suspected sepsis in the emergency room. METHODS: The study population comprised patients with suspected bacterial infection as their main diagnosis for admission to the emergency room at two University-based hospitals. Mortality within the first 28 days from admission was predicted using logistic regression with the following variables: age, immunosuppressive systemic disease, general systemic disease, Shock Index, temperature, respiratory rate, Glasgow Coma Scale score, leucocyte counts, platelet counts and creatinine. Also, with the same input and output variables, a probabilistic neural network was trained with an adaptive genetic algorithm. The network had three neurone layers: 10 neurones in the input layer, 368 in the hidden layer and two in the output layer. Calibration was measured using the Hosmer-Lemeshow goodness-of-fit test and discrimination was determined using receiver operating characteristic curves. RESULTS: A total of 533 patients were recruited and overall 28-day mortality was 19%. The factors chosen by logistic regression (with their score in parentheses) were as follows: immunosuppressive systemic disease or general systemic disease (2), respiratory rate 24-33 breaths/min (1), respiratory rate > or = 34 breaths/min (3), Glasgow Come Scale score < or = 12 (3), Shock Index > or = 1.5 (2) and temperature < 38 degrees C (2). The network included all variables and there were no significant differences in predictive ability between the approaches. The areas under the receiver operating characteristic curves were 0.7517 and 0.8782 for the logistic model and the neural network, respectively (P = 0.037). CONCLUSION: A predictive model would be an extremely useful tool in the setting of suspected sepsis in the emergency room. It could serve both as a guideline in medical decision-making and as a simple way to select or stratify patients in clinical research. Our proposed model and the specific development method -- either logistic regression or neural networks -- must be evaluated and validated in an independent population.

Adolescent↗