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 451 records · Page 25Linked to original sources
PubMed2001

Estimation of intra-cranial neural activities by means of regularized neural-network-based inversion techniques.

Artificial neural networks can be exploited to solve inverse problems arising from the estimation of neural activities in the brain. In this paper, we review the network inversion techniques for solving inverse problems with special attention directed towards electroencephalographic dipole localization and the improvement of positron emission tomography. In our regularized network inversion technique, for stabilizing the solution, we explicitly include the a priori knowledge by adding penalty terms to the energy function and/or build this knowledge into the architecture of the multi-layered neural networks that are used as an inverse problem solver. In the electroencephalogram analysis, the consensus term added to the energy function facilitated 3-dipole localization for visually evoked potentials. Effectiveness of our regularization is shown in improving the positron emission tomographic images and for generating metabolic images of the brain, under the constraints given by the a priori knowledge inherent to the measurement systems and physiological rules.

Algorithms↗
PubMed1994

Differentiation of beats of ventricular and sinus origin using a self-training neural network.

Despite advances in the computerized detection of arrhythmias, arrhythmia recognition by morphological waveform analysis still poses a difficult problem. Artificial neural networks, computer algorithms that are self-trained by an analog of biological synaptic modification to perform pattern recognition, hold great promise for the differentiation of various cardiac rhythms. The goal of this study was to differentiate beats of sinus and ventricular origin on a global basis and on a patient-specific basis by the use of artificial neural network analysis. Neural networks were trained to recognize digitized intracardiac electrograms (9 patients) and surface electrocardiograms (11 patients) obtained during sinus rhythm and ventricular tachycardia. After training, sinus rhythm or ventricular tachycardia beats were input into the neural network, and classified as to their origin. By the use of modified receiver operating characteristic curve plots, it was possible to differentiate with high sensitivity and specificity between beats of sinus origin and ventricular origin in all patients. The addition of high amounts of noise to the beats did not markedly degrade the performance of the surface ECG neural networks, and still allowed high sensitivity in differentiating beats of sinus origin from beats of ventricular origin, especially when noise was added to the training set. Neural networks provided sensitive and specific detection of cardiac electrical activity during sinus rhythm and ventricular tachycardia, and may play an important role in allowing development of improved arrhythmia recognition and management systems.

Aged↗
PubMed1991

Parallel distributed processing and neural networks: origins, methodology and cognitive functions.

Parallel Distributed Processing (PDP), a computational methodology with origins in Associationism, is used to provide empirical information regarding neurobiological systems. Recently, supercomputers have enabled neuroscientists to model brain behavior-relationships. An overview of supercomputer architecture demonstrates the advantages of parallel over serial processing. Histological data provide physical evidence of the parallel distributed nature of certain aspects of the human brain, as do corresponding computer simulations. Whereas sensory networks follow more sequential neural network pathways, in vivo brain imaging studies of attention and rudimentary language tasks appear to involve multiple cortical and subcortical areas. Controversy remains as to whether associative models or Artificial Intelligence symbolic models better reflect neural networks of cognitive functions; however, considerable interest has shifted towards associative models.

Artificial Intelligence↗
PubMed1997

Probabilistic neural network prediction of ascites in broilers based on minimally invasive physiological factors.

A Probabilistic Neural Network (PNN) was trained to predict ascites in broilers based on minimally invasive inputs (i.e., physiological factors that do not require the death of the bird). A PNN is a supervised, three-layer, artificial neural network that classifies input patterns (e.g., physiological data) into specific output categories (e.g., ascites or no ascites). The PNN inputs were O2 level in the blood, body weight, electrocardiogram (ECG), hematocrit, S wave, and heart rate of individual birds. These data were from three experiments that have been described previously (Roush et al., 1996a,b). The three data sets were pooled into a combined data set for a total of 170 observations. From the pooled data, a training set (117 birds), a calibration set (17 birds), and a verification set (36 birds) were extracted. The PNN was trained on the training data set. To prevent the PNN from overfitting the training data, the neural network was evaluated on its ability to make correct predictions of the calibration data set. At the point at which the neural network made the highest number of correct classifications for the calibration data set, the trained neural network was saved on the computer. When the PNN was applied to the complete data set, the sensitivity or proportion of the birds with ascites that the PNN correctly diagnosed was 0.97 (75/77 birds). The specificity or proportion of birds that the PNN made a correct diagnosis of not having ascites was 0.98 (91/93 birds). When the PNN was applied to the verification data set, which was not subjected to neural network training, the sensitivity was 0.95 (19/20) and the specificity was 0.88 (14/16 birds). Use of models developed with artificial neural networks may enhance the diagnosis of ascites in broilers. The results may be useful in choosing and developing broiler strains that do not have a propensity for ascites.

Animals↗
PubMed1998

Neural networks: a new technique for development of decision support systems in dentistry.

OBJECTIVES: To outline the key concepts of neural network based systems and to evaluate the potential applications of such systems in dentistry. DATA SOURCES: Published work on neural networks. CONCLUSIONS: Neural networks may initially seem complex and computer intensive, but actually integrate well with a clinical environment. Neural network expert systems may be trained with only clinical data and as such can be used where 'rule based' decision making is not possible. This is the case in many clinical situations. Neural networks may therefore become important decision making tools within dentistry and have applications both in improving clinical care and in maximizing the cost benefit of care.

Algorithms↗
PubMed1993

Neural networks for molecular sequence classification.

A neural network classification method has been developed as an alternative approach to the search/organization problem of large molecular databases. Two artificial neural systems have been implemented on a Cray supercomputer for rapid protein/nucleic acid sequence classifications. The neural networks used are three-layered, feed-forward networks that employ back-propagation learning algorithm. The molecular sequences are encoded into neural input vectors by applying an n-gram hashing method or a SVD (singular value decomposition) method. Once trained with known sequences in the molecular databases, the neural system becomes an associative memory capable of classifying unknown sequences based on the class information embedded in its neural interconnections. The protein system, which classifies proteins into PIR (Protein Identification Resource) superfamilies, showed a 82% to a close to 100% sensitivity at a speed that is about an order of magnitude faster than other search methods. The pilot nucleic acid system, which classifies ribosomal RNA sequences according to phylogenetic groups, has achieved a 100% classification accuracy. The system could be used to reduce the database search time and help organize the molecular sequence databases. The tool is generally applicable to any databases that are organized according to family relationships.

Base Sequence↗
PubMed1998

Artificial neural networks in pediatric urology: prediction of sonographic outcome following pyeloplasty.

PURPOSE: Computerized artificial neural networks are analogous to biological neuronal systems. Since they may be trained to recognize the relevance of complex patterns in data, neural networks may be useful for decision making in the multifactorial management of ureteropelvic junction obstruction. We determine the ability of a customized neural network to predict sonographic outcome after pyeloplasty in children with ureteropelvic junction obstruction. MATERIALS AND METHODS: A data set was constructed with 242 demographic, clinical, radiological and surgical elements. We analyzed the available retrospective data in 100 consecutive children who underwent unilateral pyeloplasty for ureteropelvic junction obstruction chosen from all 144 surgically treated for ureteropelvic junction obstruction between 1993 and 1995. One radiologist reviewed all film data and provided a final sonographic outcome designation in each case. We wrote a set of computer programs to construct a neural network. A composite 4-layer network was built with output nodes representing 4 possible sonographic outcomes. The 100 patient data set was randomly divided into 84 training and 16 testing examples. RESULTS: The neural network correctly predicted all 5 of 5 significantly improved, 7 of 7 improved, 2 of 2 same and 2 of 2 worse sonogram results after pyeloplasty. Therefore, sensitivity and specificity were 100% for all 4 outcomes. Linear regression analysis of the data yielded inferior sensitivity and specificity values (52 to 94%), confirming that ureteropelvic junction obstruction is a nonlinear data analysis problem. CONCLUSIONS: The 100% accuracy, sensitivity and specificity of our neural network in this pilot study provide evidence of the value of the neural computational approach for the modern exploration and modeling of the clinical problem of pediatric ureteropelvic junction obstruction.

Child↗
PubMed2004

A novel neural network for nonlinear convex programming.

In this paper, we present a neural network for solving the nonlinear convex programming problem in real time by means of the projection method. The main idea is to convert the convex programming problem into a variational inequality problem. Then a dynamical system and a convex energy function are constructed for resulting variational inequality problem. It is shown that the proposed neural network is stable in the sense of Lyapunov and can converge to an exact optimal solution of the original problem. Compared with the existing neural networks for solving the nonlinear convex programming problem, the proposed neural network has no Lipschitz condition, no adjustable parameter, and its structure is simple. The validity and transient behavior of the proposed neural network are demonstrated by some simulation results.

Neural Networks, Computer↗
PubMed1994

Artificial neural networks for early detection and diagnosis of cancer.

Why use neural networks? The reasons commonly cited in the literature for using artificial neural networks for any problem are many and varied. They learn from experience. They work where other algorithms fail. They generalize from the training examples to perform well on independent test data. They reduce the number of false alarms without increasing significantly the number of false negatives. They are fast and are easier to use than conventional statistical techniques, especially when multiple prognostic factors are needed for a given problem. These factors have been overly promoted for the neural techniques. The common theme of this paper is that artificial neural networks have proven to be an interesting and useful alternate processing strategy. Artificial neural techniques, however, are not magical solutions with mystical abilities that work without good engineering. With good understanding of their capabilities and limitations they can be applied productively to problems in early detection and diagnosis of cancer. The specific cancer applications which will be used to demonstrate current work in artificial neural networks for cancer detection and diagnosis are breast cancer, liver cancer and lung cancer.

Diagnosis, Computer-Assisted↗
PubMed1997

Neural networks as predictors of outcomes in alcoholic patients with severe liver disease.

We developed and evaluated neural networks as predictors of outcomes in alcoholic patients with severe liver disease using commonly available clinical and laboratory values. Hospital charts of 144 patients were reviewed. Nine variables (five laboratory, four clinical) were recorded along with in-hospital death or survival. Data were organized into separate development and validation sets. Neural network predictions of survival were compared with those of the Maddrey discriminant function and logistic regression models developed on the same data. Model performance was evaluated by comparing areas under receiver-operating characteristic (ROC) curves and the distributions of model scores. Survivors had significantly different laboratory and clinical characteristics, the most important being a higher prothrombin time, lower bilirubin, and lower incidence of encephalopathy. Neural network performance was significantly better than that of the Maddrey score (ROC areas, 81.5% vs. 73.8%; P = .04). The ROC area for neural networks was similar to that of logistic regression (ROC area 78.2%; P = .3), but the neural networks were more successful in classifying patients into low- and high-risk groups (P < .001). A neural network score with laboratory data from hospital-day 7 improved prognostic accuracy further to 84.3%. After adjusting for baseline risk, the neural network change in illness severity was still a significant predictor of mortality (P = .001). Neural networks using clinical and laboratory data showed a high prognostic accuracy for predicting mortality in alcoholic patients with severe liver disease.

Adult↗
PubMed1991

Using neural networks to diagnose cancer.

While artificial brains are in the realm of science fiction, artificial neural networks (ANNs) are scientific facts. An artificial neural network is a computational structure modeled somewhat on the neural structure of the brain; both have many highly interconnected processing elements. These biologically inspired processing elements are taught by feeding examples until the results are acceptable. In the past 5 years, neural networks have become successful in providing meaningful second opinions in clinical diagnosis. In our research, a prototype artificial neural network was trained on numeral ultrasound data of 52 actual cases and then correctly identified renal cell carcinoma from renal cysts and other conditions without diagnostic errors. Our nonlinear artificial neural network was trained on software using the standard backpropagation paradigm on a 80386 microcomputer. Our ANN learned from ultrasound data in 52 cases (17 malignant, 30 cysts, and 5 other) at a Memphis hospital. The trained prototype performed without error on 47 cases which were not in the data used for training. This prototype must be validated by extending this study to more cases.

Carcinoma, Renal Cell↗
PubMed1998

State space neural network. Properties and application.

In this paper, a specific neural network based model for the identification of non-linear systems is proposed. This neural network structure is able to identify a state space non-linear model of the plant. The use of the state space representation presents several advantages that must be taken into account. One of the most important advantages is that the resulting neural model can be easily linearized around different operating points, allowing application of classical stability theorems from the linear systems domain to this class of neural networks. In this way, some useful theoretical results for neural modelling and identification have been obtained and presented in the paper. In this paper, several stability theorems and practical implementation issues are addressed. Examples are also presented which show the training capability of the neural network and the validity of the theory presented.

Journal Article↗
PubMed1993

Artificial neural networks for predicting failure to survive following in-hospital cardiopulmonary resuscitation.

BACKGROUND: Neural networks are an artificial intelligence technique that uses a set of nonlinear equations to mimic the neuronal connections of biological systems. They have been shown to be useful for pattern recognition and outcome prediction applications, and have the potential to bring artificial intelligence techniques to the personal computers of practicing physicians, assisting them with a variety of medical decisions. It is proposed that such an artificial neural network can be trained, using information available at the time of admission to the hospital, to predict failure to survive following in-hospital cardiopulmonary resuscitation (CPR). METHODS: The age, sex, heart rate, and 21 other clinical variables were collected on a consecutive series of 218 adult patients undergoing CPR at a 295-bed public acute-care hospital. The data set was divided into two groups. A neural network was trained to predict failure to survive to discharge following CPR, using one group as the training set and the other as the testing set. The procedure was then reversed, and the results of the two networks were combined to form an aggregate network. RESULTS: The trained aggregate neural network had a sensitivity of 52.1% and a positive predictive value of 97% for the prediction of failure to survive following CPR. The relative risk of actually failing to survive to discharge following CPR for a patient predicted not to survive was 11.3 (95% CI 3.3 to 38.2). CONCLUSIONS: Predicting failure to survive following CPR is but one possible application of neural network technology. It demonstrates how this technique can assist physicians in medical decision making. Future work should attempt to improve the positive predictive value of the neural network, to consider combining it with an expert system, and to compare it with other predictive tools. Once validated, the network can be distributed as a separate application for use by practicing physicians.

Adolescent↗
PubMed2004

Neural networks for nonlinear and mixed complementarity problems and their applications.

This paper presents two feedback neural networks for solving a nonlinear and mixed complementarity problem. The first feedback neural network is designed to solve the strictly monotone problem. This one has no parameter and possesses a very simple structure for implementation in hardware. Based on a new idea, the second feedback neural network for solving the monotone problem is constructed by using the first one as a subnetwork. This feedback neural network has the least number of state variables. The stability of a solution of the problem is proved. When the problem is strictly monotone, the unique solution is uniformly and asymptotically stable in the large. When the problem has many solutions, it is guaranteed that, for any initial point, the trajectory of the network does converge to an exact solution of the problem. Feasibility and efficiency of the proposed neural networks are supported by simulation experiments. Moreover, the feedback neural network can also be applied to solve general nonlinear convex programming and nonlinear monotone variational inequalities problems with convex constraints.

Feedback↗
PubMed2005

Artificial neural networks as a useful tool to predict the risk level of Betula pollen in the air.

An increasing percentage of the European population suffers from allergies to pollen. The study of the evolution of air pollen concentration supplies prior knowledge of the levels of pollen in the air, which can be useful for the prevention and treatment of allergic symptoms, and the management of medical resources. The symptoms of Betula pollinosis can be associated with certain levels of pollen in the air. The aim of this study was to predict the risk of the concentration of pollen exceeding a given level, using previous pollen and meteorological information, by applying neural network techniques. Neural networks are a widespread statistical tool useful for the study of problems associated with complex or poorly understood phenomena. The binary response variable associated with each level requires a careful selection of the neural network and the error function associated with the learning algorithm used during the training phase. The performance of the neural network with the validation set showed that the risk of the pollen level exceeding a certain threshold can be successfully forecasted using artificial neural networks. This prediction tool may be implemented to create an automatic system that forecasts the risk of suffering allergic symptoms.

Air Pollutants↗
PubMed2004

A high performance k-NN approach using binary neural networks.

This paper evaluates a novel k-nearest neighbour (k-NN) classifier built from binary neural networks. The binary neural approach uses robust encoding to map standard ordinal, categorical and numeric data sets onto a binary neural network. The binary neural network uses high speed pattern matching to recall a candidate set of matching records, which are then processed by a conventional k-NN approach to determine the k-best matches. We compare various configurations of the binary approach to a conventional approach for memory overheads, training speed, retrieval speed and retrieval accuracy. We demonstrate the superior performance with respect to speed and memory requirements of the binary approach compared to the standard approach and we pinpoint the optimal configurations.

Algorithms↗
PubMed1998

Application of artificial neural networks to eating disorders.

An experimental application of Artificial Neural Networks to Eating Disorders is presented. The sample, composed of 172 cases (all women) collected at the Centre for the Diagnosis and Treatment of Eating Disorders of the 1st Medical Division of the St. Eugenio Hospital of Rome, was subdivided, on the basis of the diagnosis made by the specialist of the St. Eugenio, into four classes: Anorexia Nervosa (AN), Nervous Bulimia (NB), Binge Eating Disorders (BED) and Psychogenic Eating Disorders that are Not Otherwise Specified (PED-NOS). The data base was composed of 124 different variables: generic information, alimentary behavior, eventual treatment and hospitalization, substance use, menstrual cycles, weight and height, hematochemical and instrumental examinations, psychodiagnostic tests, etc. The goal of this experiment was to verify the accuracy of the Neural Networks in recognising anorexic and bulimic patients. This article describes 6 experiments, using a Feed Forward Neural Network, each one using different variables. Starting from only the generic variables (life styles, family environment, etc.) and hematoclinical and instrumental examinations, a Neural Networks provided 86.94% of the prediction precision. This work is meant to be a first contribution to creating diagnostic procedures for Eating Disorders, that would be simple and easy-to-use by professionals who are neither psychologists nor psychiatrists nor psychotherapists but who are, however, among the first to meet these patients and who are therefore called upon to give such patients the very first pieces of advice on seeking proper treatment.

Anorexia Nervosa↗
PubMed2001

A new algorithm to design compact two-hidden-layer artificial neural networks.

This paper describes the cascade neural network design algorithm (CNNDA), a new algorithm for designing compact, two-hidden-layer artificial neural networks (ANNs). This algorithm determines an ANN's architecture with connection weights automatically. The design strategy used in the CNNDA was intended to optimize both the generalization ability and the training time of ANNs. In order to improve the generalization ability, the CNDDA uses a combination of constructive and pruning algorithms and bounded fan-ins of the hidden nodes. A new training approach, by which the input weights of a hidden node are temporarily frozen when its output does not change much after a few successive training cycles, was used in the CNNDA for reducing the computational cost and the training time. The CNNDA was tested on several benchmarks including the cancer, diabetes and character-recognition problems in ANNs. The experimental results show that the CNNDA can produce compact ANNs with good generalization ability and short training time in comparison with other algorithms.

Algorithms↗