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 901 records · Page 50Linked to original sources
PubMed1999

Preliminary study on application of artificial neural network to the diagnosis of Alzheimer's disease with magnetic resonance imaging.

OBJECTIVE: Artificial neural network is first used in the measurement study of brain of Alzheimer's disease using MRI, and a completely new pattern discriminating method is adopted, so as to take advantage of MRI to diagnose and identify AD patients. METHODS: 12 patients with probable AD (aged 65.33 +/- 8.62 years) and 36 normal controls matched with age and gender (aged 65.81 +/- 74.37 years) were studied. MRI are performed on Siemens Magnetom IMPACT 1.0 T; eight interesting brain structures including sixteen regions (left and right) indices are measured and studied; SPSS software and BP network software made by authors respectively were used to process and analyze the measured data. RESULTS: Using artificial neural network to the same regions and data, both the sensitivity and accuracy were found higher than using the traditional discrimination function analysis method; the indices of amygdala, hippocampus, parahippocampal gyrus, temporal lobe, and temporal horn, these five structures could completely differentiate AD from normal controls; new cases were successfully diagnosed. CONCLUSIONS: Artificial neural network combining with MRI is probable to become a useful and reliable clinical tool to diagnose AD patients.

Aged↗
PubMed2002

Prediction of human skin permeability using a combination of molecular orbital calculations and artificial neural network.

This study was carried out to develop a novel method for predicting the skin permeability coefficient (log K(p)) of compounds from their three-dimensional molecular structure using a combination of molecular orbital (MO) calculation and artificial neural network. Human skin permeability data on 92 structurally diverse compounds were analyzed. The molecular descriptors of each compound, such as the dipole moment, polarizability, sum of charges of nitrogen and oxygen atoms (sum(N,O)), and sum of charges of hydrogen atoms bonding to nitrogen or oxygen atoms (sum(H)) were obtained from MO calculations. The correlation between these molecular descriptors and log K(p) was examined using feed-forward back-propagation neural networks. To improve the generalization capability of a neural network, the network was trained with input patterns given 5% random noise. The neural network model with a configuration of 4-4-1 for input, hidden, and output layers was much superior to the conventional multiple linear regression model in terms of root mean square (RMS) errors (0.528 vs. 0.930). A "leave-one-out" cross-validation revealed that the neural network model could predict skin permeability with a reasonable accuracy (predictive RMS error of 0.669).

Humans↗
PubMed2001

Predicting outcomes of trials of labor in women attempting vaginal birth after cesarean delivery: a comparison of multivariate methods with neural networks.

OBJECTIVE: Our aim was to assess the utility and effectiveness of a neural network for predicting the likelihood of success of a trial of labor, relative to standard multivariate predictive models. STUDY DESIGN: We identified 100 failed trials of labor and 300 successful trials of labor in women with a prior cesarean delivery performed at our institution. Information was collected on >70 potential predictors of labor outcomes from the medical records, including demographic, historical, and past obstetric information, as well as information from the index pregnancy. Bivariate analyses comparing women in whom a trial of labor failed with those whose trial succeeded were performed. These initial analyses were used to select variables for inclusion into our muitivariate predictive model. From the same data we trained and tested a neural network, using a back-propagation algorithm. The test characteristics of the multivariate predictive model and the neural network were compared. RESULTS: From the bivariate analysis a history of substance abuse (adjusted odds ratio, 0.27; 95% confidence interval, 0.09-0.80), a successful prior vaginal birth after cesarean delivery (adjusted odds ratio, 0.13; 95% confidence interval, 0.05-0.31), cervical dilatation at admission (adjusted odds ratio, 0.53; 95% confidence interval, 0.31-0.88), and the need for labor augmentation (adjusted odds ratio, 2.15; 95% confidence interval, 1.14-4.06) were ultimately discovered to be important in predicting the likelihood of the success or failure of a trial of labor. With these variables in the predictive model the sensitivity of the derived rule for predicting failure was 77%, the specificity was 65%, and the overall accuracy was 69%. We also built a network using the 4 variables that were included in the final multivariate model. We were unable to achieve the same degree of sensitivity and specificity that we observed with the regression-based predictive model (sensitivity and specificity, 59% and 44%). CONCLUSION: In this study a standard multivariate model was better able to predict outcome in women ttempting a trial of labor.

Adult↗
PubMed2001

Establishing an index arbitrage model by applying neural networks method--a case study of Nikkei 225 index.

This paper applies the neural network method to establish an index arbitrage model and compares the arbitrage performances to that from traditional cost of carry arbitrage model. From the empirical results of the Nikkei 225 stock index market, following conclusions can be stated: (1) The basis will get enlarged for a time period, more profitability may be obtained from the trend. (2) If the neural network is applied within the index arbitrage model, twofold of return would be obtained than traditional arbitrage model can do. (3) If the T_basis has volatile trend, the neural network arbitrage model will ignore the peak. Although arbitrageur would lose the chance to get profit, they may reduce the market impact risk.

Computer Simulation↗
PubMed1998

Prediction and cross-validation of neural networks versus logistic regression: using hepatic disorders as an example.

The authors developed and cross-validated prediction models for newly diagnosed cases of liver disorders by using logistic regression and neural networks. Computerized files of health care encounters from the Fallon Community Health Plan were used to identify 1,674 subjects who had had liver-related health services between July 1, 1992, and June 30, 1993. A total of 219 subjects were confirmed by review of medical records as incident cases. The 1,674 subjects were randomly and evenly divided into training and test sets. The training set was used to derive prediction algorithms based solely on the automated data; the test set was used for cross-validation. The area under the Receiver Operating Characteristic curve for a neural network model was significantly larger than that for logistic regression in the training set (p = 0.04). However, the performance was statistically equivalent in the test set (p = 0.45). Despite its superior performance in the training set, the generalizability of the neural network model is limited. Logistic regression may therefore be preferred over neural network on the basis of its established advantages. More generalizable modeling techniques for neural networks may be necessary before they are practical for medical research.

Algorithms↗
PubMed2001

The determination of cardiac surgical risk using artificial neural networks.

BACKGROUND: In the study presented here, an artificial neural network was used to "learn" the relationship between 11 risk factors and patient surgical outcome (survival or death). The network was then used to predict the surgical outcome of other patients. METHODS: Eleven risk factors were presented as inputs to an artificial neural network (ANN). The ANN model was developed by training and testing on 1875 patients. The results of the ANN were compared with the results from two versions of the VA surgical risk model (Denver model). Cutoffs were determined for each model in order to compare their results. RESULTS: The ANN model gave the best results when compared with the new and old Denver models. The ANN model had the lowest overall percentage of error. It predicted living patients with an error of 14% and death with an error of approximately 31.0%. The old Denver model predicted living patients with an error of 15% and deaths with an error of 31%. The new Denver model predicts living patients with an error around 18% and the deaths with an error of 31%. CONCLUSIONS: The combined predictions of the ANN model were slightly more accurate than either the new or the old Denver models. The ANN model was created from 1875 patients in about 1 month, while both of the Denver models were developed over a 3 year time period and used more than 12,000 patients. Additionally, the ANN model is easily modified, allowing instant addition or deletion of parameters to suit the users needs.

Cardiac Surgical Procedures↗
PubMed1994

A fuzzy neural network for knowledge learning.

This paper presents a fuzzy neural network for learning the knowledge of a fuzzy logic rule-based system. The network contains five layers: an Input Layer, Membership-function Layer, AND Layer, OR Layer, and Defuzzification Layer. We propose a backpropagation-like learning algorithm to train this neural network to acquire the fuzzy rules and to fine-tune the knowledge on the parameters of AND and OR nodes. Compared with methods other than the gradient descent search, the proposed learning process acquires more precise knowledge. In addition, the functions of the AND and OR nodes in the network are formulated with the minimum or maximum operations, respectively. Therefore, the adjustments of the learnable weights (parameters) can be focused on the dominant terms related to the (minimum/maximum) operations. The convergence time for the proposed learning algorithm is much faster than that for conventional backpropagation algorithms. In summary, the learnable weights (parameters) of the network are adjusted very quickly to obtain precise knowledge. Simulation results show that in learning the truck backer-upper problem, our network completes the training procedure in only several dozen epochs with an error rate of less than 1%.

Algorithms↗
PubMed1994

Classification of chromosomes using a probabilistic neural network.

This paper describes the application of a probabilistic neural network (PNN) to the classification of normal human chromosomes. The inputs to the network are 30 different features extracted from each chromosome in digitized images of metaphase spreads. The output is 1 of 24 different classes of chromosomes (the 22 autosomes plus the sex chromosomes X and Y). An updating procedure was implemented to take advantage of the fact that in a normal somatic cell only two chromosomes can be assigned to each class. The network has been tested using the Copenhagen, Edinburgh, and Philadelphia databases of digitized images of human chromosomes. The recognition rates achieved in this study are superior to those reported using either the maximum likelihood or back propagation neural network techniques.

Chromosomes, Human↗
PubMed1993

Neural network analysis of ventilation-perfusion lung scans.

A neural network model was constructed to interpret ventilation-perfusion (V/Q) lung scans. This model was trained with data from 100 consecutive V/Q scans with pulmonary angiographic correlation. The network was constructed from 28 input parameters that described various standard V/Q findings, which were fed into a single hidden layer that contained 10-20 nodes. The network output indicated the percentage probability of pulmonary embolism for each set of findings on V/Q scans. This network was then used to classify 28 new scans; the resultant classifications were compared with the rankings of an experienced observer who read the scans without knowledge of the correlative angiographic data. The network with 15 hidden nodes outperformed the experienced observer in prediction of the likelihood of pulmonary embolism in the 28-case test set (P = .039). The neural network has several advantages over current algorithms for interpretation of V/Q scans, including the ability to synthesize many variables into a single conclusion and to learn, or modify itself, at exposure to additional data.

Adult↗
PubMed1993

Artificial neural networks in mammography: application to decision making in the diagnosis of breast cancer.

The authors investigated the potential utility of artificial neural networks as a decision-making aid to radiologists in the analysis of mammographic data. Three-layer, feed-forward neural networks with a back-propagation algorithm were trained for the interpretation of mammograms on the basis of features extracted from mammograms by experienced radiologists. A network that used 43 image features performed well in distinguishing between benign and malignant lesions, yielding a value of 0.95 for the area under the receiver operating characteristic curve for textbook cases in a test with the round-robin method. With clinical cases, the performance of a neural network in merging 14 radiologist-extracted features of lesions to distinguish between benign and malignant lesions was found to be higher than the average performance of attending and resident radiologists alone (without the aid of a neural network). The authors conclude that such networks may provide a potentially useful tool in the mammographic decision-making task of distinguishing between benign and malignant lesions.

Female↗
PubMed1997

Latency estimation of auditory brainstem response by neural networks.

In the clinical application of auditory brainstem responses (ABRs), the latencies of five to seven main peaks are extremely important parameters for diagnosis. In practice, the latencies have mainly been done by manual measurement so far. In recent years, some new techniques have been developed involving automatic computer recognition. Computer recognition is difficult, however, since some peaks are complicated and vary a lot individually. In this paper, we introduce an artificial neural network method for ABR research. The detection of ABR is performed by using artificial neural networks. A proper bandpass filter is designed for peak extraction. Moreover, a new approach to estimate the latencies of the peaks by artificial neural networks is presented. The neural networks are studied in relation to the selection of model, number of layers and number of neurons in each hidden layer. Experimental results are described showing that artificial neural networks are a promising method in the study of ABR.

Evoked Potentials, Auditory, Brain Stem↗
PubMed1988

Neural networks counting chimes.

It is shown that the ideas that led to neural networks capable of recalling associatively and asynchronously temporal sequences of patterns can be extended to produce a neural network that automatically counts the cardinal number in a sequence of identical external stimuli. The network is explicitly constructed, analyzed, and simulated. Such a network may account for the cognitive effect of the automatic counting of chimes to tell the hour. A more general implication is that different electrophysiological responses to identical stimuli, at certain stages of cortical processing, do not necessarily imply synaptic modification, a la Hebb. Such differences may arise from the fact that consecutive identical inputs find the network in different stages of an active temporal sequence of cognitive states. These types of networks are then situated within a program for the study of cognition, which assigns the detection of meaning as the primary role of attractor neural networks rather than computation, in contrast to the parallel distributed processing attitude to the connectionist project. This interpretation is free of homunculus, as well as from the criticism raised against the cognitive model of symbol manipulation. Computation is then identified as the syntax of temporal sequences of quasi-attractors.

Computer Simulation↗
PubMed2000

[The medium optimization of xylitol fermentation based on neural networks and genetic algorithms].

Using genetic algorithms(GA) for medium optimization of xylitol fermentation, and coupling neural networks model for predicting xylitol concentration is introduced. The medium compose determined by GA is as input data of the neural networks, while the output data predicted by neural networks is as suitable value of GA for predicting. The optimum medium is further validated by experimentation. The good result, which save the experimental workload and charge, enhance the level of xylitol fermentation as well as reduced the medium consume is obtained.

Algorithms↗
PubMed2004

Entropy-based generation of supervised neural networks for classification of structured patterns.

Sperduti and Starita proposed a new type of neural network which consists of generalized recursive neurons for classification of structures. In this paper, we propose an entropy-based approach for constructing such neural networks for classification of acyclic structured patterns. Given a classification problem, the architecture, i.e., the number of hidden layers and the number of neurons in each hidden layer, and all the values of the link weights associated with the corresponding neural network are automatically determined. Experimental results have shown that the networks constructed by our method can have a better performance, with respect to network size, learning speed, or recognition accuracy, than the networks obtained by other methods.

Entropy↗
PubMed1999

Neural network prediction of obstructive sleep apnea from clinical criteria.

STUDY OBJECTIVES: Clinical prediction models for the diagnosis of obstructive sleep apnea (OSA) have lacked the accuracy necessary to confidently replace polysomnography (PSG). Artificial neural networks are computer programs that can be trained to predict outcomes based on experience. This study was conducted to test the hypothesis that a generalized regression neural network (GRNN) could accurately classify patients with OSA from clinical data. STUDY DESIGN: Retrospective review. SETTING: Regional sleep referral center. PATIENTS: Randomly selected records of patients referred for possible OSA. MEASUREMENTS: The neural network was trained using 23 clinical variables from 255 patients, and the predictive performance was evaluated using 150 other patients. RESULTS: The prevalence of OSA in this series of 405 patients (293 men and 112 women) was 69%. The trained GRNN had an accuracy of 91.3% (95% confidence interval [CI], 86.8 to 95.8). The sensitivity was 98.9% for having OSA (95% CI, 96.7 to 100), and the specificity was 80% (95% CI, 70 to 90). The positive predictive value that the patient would have OSA was 88.1% (95% CI, 81.8 to 94.4), whereas the negative predictive value that the patient would not have OSA (if so classified) was 98% (95% CI, 94 to 100). CONCLUSIONS: Appropriately trained GRNN has the ability to accurately rule in OSA from clinical data, and GRNN did not misclassify patients with moderate to severe OSA. In this study, use of the neural network could have reduced the number of PSG studies performed. Prospective validation of the neural network for the diagnosis of OSA is now required.

Adolescent↗
PubMed2003

On-line lower-order modeling via neural networks.

This paper presents a novel method to determine the parameters of a first-order plus dead-time model using neural networks. The outputs of the neural networks are the gain, dominant time constant, and apparent time delay. By combining this algorithm with a conventional PI or PID controller, we also present an adaptive controller which requires very little a priori knowledge about the plant under control. The simplicity of the scheme for real-time control provides a new approach for implementing neural network applications for a variety of on-line industrial control problems. Simulation and experimental results demonstrate the feasibility and adaptive property of the proposed scheme.

Algorithms↗
PubMed2000

Flow injection analysis of fluoride: optimization of experimental conditions and non-linear calibration using artificial neural networks.

This paper deals with the application of artificial neural networks (ANNs) to two common problems in spectroscopy: optimization of experimental conditions and non-linear calibration of the result, with particular reference to the determination of fluoride by flow injection analysis (FIA). The FIA system was based on the formation of a blue ternary complex between zirconium(IV), p-methyldibromoarsenazo and F- with the maximum absorption wavelength at 635 nm. First, optimization in terms of sensitivity and sampling rate was carried out by using jointly a central composite design and ANNs, and a neural network with a 3-7-1 structure was confirmed to be able to provide the maximum performance. Second, the relationship between the concentration of fluoride and its absorbance was modeled by ANNs. In this process, cross-validation and leave-k-out were used. The results showed that good prediction was attained in the 1-4-1 neural net. The trained networks proved to be very powerful in both applications. The proposed method was successfully applied to the determination of free fluoride in tea and toothpaste with recoveries between 96 and 101%.

Calibration↗
PubMed1997

Neural network modeling accurately predicts the functional outcome of stroke survivors with moderate disabilities.

OBJECTIVE: To predict the place of discharge or discharge Functional Independence Measure (FIM) score for stroke survivors with moderate disability using neural network modeling. Our previous work demonstrated that the FIM predicts the level of recovery for stroke survivors with either severe or mild disabilities. DESIGN: Neural network analysis. SETTING: Tertiary care rehabilitation program. PATIENTS: One hundred forty-seven consecutive stroke survivors admitted for rehabilitation with admission FIM scores between 37 and 96 were used as the training and internal test set. Seventeen other randomly selected stroke survivors were used as the external test set. INTERVENTION: A neural network model was developed using a small set of clinical variables and the admission FIM score. MAIN OUTCOME MEASURE: Neural network model predicting place of discharge or discharge FIM score. RESULTS: A working and accurate model was developed to predict the discharge FIM score. The model was able to predict the 17 external test cases with an accuracy = 88%, sensitivity = 83%, specificity = 91%, positive predictive value = 83%, and negative predictive value = 91%. CONCLUSION: Neural network modeling is useful in the prediction of functional recovery and helps in discharge planning and allocation of rehabilitation resources.

Aged↗