Multicomponent kinetic determinations using artificial neural networks

Neural networks were successfully used for multicomponent kinetic determinations of species with rate constant ratios approaching unity without the aid of spectral discrimination. The ensuing method relies on two inputs describing the profile of the kinetic curve for each mixture, which is obtained by preprocessing kinetic data using nonlinear least-squares regression. A straightforward network architecture (2:4s:21) was used to resolve mixtures of 2- and 3-chlorophenol; the trained network estimated the concentrations of both components in the mixture with a relative standard error of prediction of approximately 5%, which is much lower than that obtained with Kalman filtering. The effect of some variables such as the rate constant and analyte concentration ratios on the proposed multicomponent determination is discussed.     

Simulation of complex movements using artificial neural networks

A simulated network for controlling a six-legged, insect-like walking system is proposed. The network contains internal recurrent connections, but important recurrent connections utilize the loop through the environment. This approach leads to a subnet for controlling the three joints of a leg during its swing which is arguably the simplest possible solution. The task for the stance subnet appears more difficult because the movements of a larger and varying number of joints (9-18: three for each leg in stance) have to be controlled such that each leg contributes efficiently to support and propulsion and legs do not work at cross purposes. Already inherently non-linear, this task is further complicated by four factors: 1) the combination of legs in stance varies continuously. 2) during curve walking, legs must move at different speeds, 3) on compliant substrates, the speed of the individual legs may vary unpredictably, and 4) the geometry of the system may vary through growth and injury or due to non-rigid suspension of the joints. This task appears to require some kind of "motor intelligence". We show that an extremely decentralized, simple controller, based on a combination of negative and positive feedback at the joint level, copes with all these problems by exploiting the physical properties of the system.     

Peptide design by artificial neural networks and computer-based evolutionary search

A technique for systematic peptide variation by a combination of rational and evolutionary approaches is presented. The design scheme consists of five consecutive steps: (i) identification of a "seed peptide" with a desired activity, (ii) generation of variants selected from a physicochemical space around the seed peptide, (iii) synthesis and testing of this biased library, (iv) modeling of a quantitative sequence-activity relationship by an artificial neural network, and (v) de novo design by a computer-based evolutionary search in sequence space using the trained neural network as the fitness function. This strategy was successfully applied to the identification of novel peptides that fully prevent the positive chronotropic effect of anti-beta1-adrenoreceptor autoantibodies from the serum of patients with dilated cardiomyopathy. The seed peptide, comprising 10 residues, was derived by epitope mapping from an extracellular loop of human beta1-adrenoreceptor. A set of 90 peptides was synthesized and tested to provide training data for neural network development. De novo design revealed peptides with desired activities that do not match the seed peptide sequence. These results demonstrate that computer-based evolutionary searches can generate novel peptides with substantial biological activity.     

Evaluation of complications of kidney transplantation using artificial neural networks

The aim of this study was to develop an artificial neural network (ANN) to differentiate between rejection, acute tubular necrosis (ATN) and normally functioning kidneys in a group of patients with renal transplants. The performance of ANN was compared with that of an experienced observer using a database of 35 patients' records, each of which included 12 quantitative parameters derived from renograms and clinical data as well as a clinical evaluation. These findings were encoded as features for a three-layered neural network to predict the outcome of biopsy or clinical diagnosis. The network was trained and tested using the jackknife method and its performance was then compared to that of a radiologist. The network was able to correctly classify 31 of the 35 original cases and gave a better diagnostic accuracy (88%) than the radiologist (83%), by showing an association between the quantitative data and the corresponding pathological results (r = 0.78, P < 0.001). We conclude that an ANN can be trained to differentiate rejection from acute tubular necrosis, as well as normally functioning transplants, with a reasonable degree of accuracy.     

Automated interpretation of myocardial SPECT perfusion images using artificial neural networks

The purpose of this study was to develop a computer-based method for automatic detection and localization of coronary artery disease (CAD) in myocardial bull's-eye scintigrams. METHODS: A population of 135 patients who had undergone both myocardial 99mTc-sestamibi rest-stress scintigraphy and coronary angiography within 3 mo was studied. Different image data reduction methods, including pixel averaging and two-dimensional Fourier transform, were applied to the bull's-eye scintigrams. After a quantitative and qualitative evaluation of these methods, 30 Fourier components were chosen as inputs to multilayer perceptron artificial neural networks. The networks were trained to detect CAD in two vascular territories, using coronary angiography as gold standard. A "leave one out" procedure was used for training and evaluation. The performance of the networks was compared to those of two human experts. RESULTS: One of the human experts detected CAD in one of two vascular territories, with a sensitivity of 54.4% at a specificity of 70.5%. The sensitivity of the networks was significantly higher at that level of specificity (77.2%, p = 0.0022). The other expert had a sensitivity of 63.2% at a specificity of 61.5%. The networks had a sensitivity of 77.2% (p = 0.038) at this specificity level as well. The differences in sensitivity between human experts and networks for the other vascular territory were all less than 6% and were not statistically significant. CONCLUSION: Artificial neural networks can detect CAD in myocardial bull's-eye scintigrams with such a high accuracy that the application of neural networks as clinical decision support tools appears to have significant potential.     

A medical expert system approach using artificial neural networks for standardized treatment planning

PURPOSE: Many radiotherapy treatment plans involve some level of standardization (e.g., in terms of beam ballistics, collimator settings, and wedge angles), which is determined primarily by tumor site and stage. If patient-to-patient variations in the size and shape of relevant anatomical structures for a given treatment site are adequately sampled, then it would seem possible to develop a general method for automatically mapping individual patient anatomy to a corresponding set of treatment variables. A medical expert system approach to standardized treatment planning was developed that should lead to improved planning efficiency and consistency. METHODS AND MATERIALS: The expert system was designed to specify treatment variables for new patients based upon a set of templates (a database of treatment plans for previous patients) and a similarity metric for determining the goodness of fit between the relevant anatomy of new patients and patients in the database. A set of artificial neural networks was used to optimize the treatment variables to the individual patient. A simplified example, a four-field box technique for prostate treatments based upon a single external contour, was used to test the viability of the approach. RESULTS: For a group of new prostate patients, treatment variables specified by the expert system were compared to treatment variables chosen by the dosimetrists. Performance criteria included dose uniformity within the target region and dose to surrounding critical organs. For this standardized prostate technique, a database consisting of approximately 75 patient records was required for the expert system performance to approach that of the dosimetrists. CONCLUSIONS: An expert system approach to standardized treatment planning has the potential of improving the overall efficiency of the planning process by reducing the number of iterations required to generate an optimized dose distribution, and to function most effectively, should be closely integrated with a dosimetric based treatment planning system.     

A new method to estimate parameters of linear compartmental models using artificial neural networks

At present, the preferred tool for parameter estimation in compartmental analysis is an iterative procedure; weighted nonlinear regression. For a large number of applications, observed data can be fitted to sums of exponentials whose parameters are directly related to the rate constants/coefficients of the compartmental models. Since weighted nonlinear regression often has to be repeated for many different data sets, the process of fitting data from compartmental systems can be very time consuming. Furthermore the minimization routine often converges to a local (as opposed to global) minimum. In this paper, we examine the possibility of using artificial neural networks instead of weighted nonlinear regression in order to estimate model parameters. We train simple feed-forward neural networks to produce as outputs the parameter values of a given model when kinetic data are fed to the networks' input layer. The artificial neural networks produce unbiased estimates and are orders of magnitude faster than regression algorithms. At noise levels typical of many real applications, the neural networks are found to produce lower variance estimates than weighted nonlinear regression in the estimation of parameters from mono- and biexponential models. These results are primarily due to the inability of weighted nonlinear regression to converge. These results establish that artificial neural networks are powerful tools for estimating parameters for simple compartmental models.     

A hybrid method to cluster protein sequences based on statistics and artificial neural networks

We have recently proposed a method, based on artificial neural networks (ANNs) to cluster protein sequences into families according to their degree of sequence similarity. The network was trained with an unsupervised learning algorithm, using, as inputs, matrix patterns derived from the bipeptide composition of the protein sequences. We describe here some further improvements to that approach. First, we propose a statistical method to cluster a set of bipeptidic matrices into families. It consists of three stages: (i) principal component analysis, (ii) determination of the optimal number M of clusters and (iii) final classification of the bipeptidic matrices into M clusters. Using a set of 444 protein sequences, we show that the classification given by the statistical method is in agreement with biological knowledge. We also show that the resulting classification is very similar to the one previously obtained with the ANN approach. Finally, we propose a new hybrid method of the statistical and ANN approaches, in which the results of the statistical method are used to choose the number of neurons and inputs of the network. We show that a network built in this way, and fed with a few principal components of the set of bipeptidic matrices as input signals, can be trained in an extremely short computing time. The resulting topological maps do not essentially differ from the ones obtained with the initial ANN approach.     

Prediction of shape defects over length of cold rolled sheet using artificial neural networks

Abstract

A model based on an artificial neural network (ANN) has been developed for prediction of flatness of cold rolled (CR) sheet in a tandem cold rolling mill for white goods applications. Various process parameters including roll bending, roll shifting, tensions between stands etc., which affect flatness of CR sheet are considered in the model. Substantial amounts of data are obtained from level II automation of PL-TCM of TATA Steel to develop the prediction model. The developed ANN model, based on back propagation algorithm, is able to predict the flatness defects like edge buckles, centre buckles for a given set of rolling parameters. The model involves a large number of process parameters and application of ANN to such kind of problems is successfully demonstrated in the present study. The model is in good agreement with the observed flatness values at different locations across the width. High coefficient of determination close to 0·919 is achieved for the prediction of flatness at edges.     

Prediction of the measured temperature after the last finishing stand using artificial neural networks

In this report the development of an artificial neural network, capable of predicting the temperature after the last finishing stand of a hot strip mill for a certain class of steels, is described. Three neural networks with different numbers of hidden nodes (3, 5 and 7) were trained. The relative standard deviation in finish temperature as predicted by the best performing neural network model (7 hidden nodes) was just over 25% smaller than that of the linear Hoogovens model. This improved accuracy can be explained by the incorrect assumption in the Hoogovens model of linear dependence of the finishing temperature on some input parameters. With the trained neural network, the influence of the various input parameters on the finishing temperature could be examined. The dependencies predicted by the neural network can be approximated by a linear fit and are a factor 2 lower for all input parameters. It is conceivable that operation of the mill using an artificial neural network for the prediction of the finishing temperature would have resulted in smaller operational fluctuations.     

Hidden neural networks

A general framework for hybrids of hidden Markov models (HMMs) and neural networks (NNs) called hidden neural networks (HNNs) is described. The article begins by reviewing standard HMMs and estimation by conditional maximum likelihood, which is used by the HNN. In the HNN, the usual HMM probability parameters are replaced by the outputs of state-specific neural networks. As opposed to many other hybrids, the HNN is normalized globally and therefore has a valid probabilistic interpretation. All parameters in the HNN are estimated simultaneously according to the discriminative conditional maximum likelihood criterion. The HNN can be viewed as an undirected probabilistic independence network (a graphical model), where the neural networks provide a compact representation of the clique functions. An evaluation of the HNN on the task of recognizing broad phoneme classes in the TIMIT database shows clear performance gains compared to standard HMMs tested on the same task.     

Agreement between artificial neural networks and experienced electrocardiographer on electrocardiographic diagnosis of healed myocardial infarction

OBJECTIVES: The purpose of this study was to compare the diagnoses of healed myocardial infarction made from the 12-lead electrocardiogram (ECG) by artificial neural networks and an experienced electrocardiographer. BACKGROUND: Artificial neural networks have proved of value in pattern recognition tasks. Studies of their utility in ECG interpretation have shown performance exceeding that of conventional ECG interpretation programs. The latter present verbal statements, often with an indication of the likelihood for a certain diagnosis, such as "possible left ventricular hypertrophy." A neural network presents its output as a numeric value between 0 and 1; however, these values can be interpreted as Bayesian probabilities. METHODS: The study was based on 351 healthy volunteers and 1,313 patients with a history of chest pain who had undergone diagnostic cardiac catheterization. A 12-lead ECG was recorded in each subject. An expert electrocardiographer classified the ECGs in five different groups by estimating the probability of anterior myocardial infarction. Artificial neural networks were trained and tested to diagnose anterior myocardial infarction. The network outputs were divided into five groups by using the output values and four thresholds between 0 and 1. RESULTS: The neural networks diagnosed healed anterior myocardial infarctions at high levels of sensitivity and specificity. The network outputs were transformed to verbal statements, and the agreement between these probability estimates and those of an expert electrocardiographer was high. CONCLUSIONS: Artificial neural networks can be of value in automated interpretation of ECGs in the near future.     

硅钢退火网络型专家系统

介绍一种用于硅钢退火工艺的网络型专家系统。该系统将人工神经网络、分类模式识别技术和有关冶金物理化学知识相结合 ,建立了优化数学模型 ,以C语言和多媒体技术实现程序设计。现场生产运行表明该系统取得了预期效果。     

Ground reaction forces in horses, assessed from hoof wall deformation using artificial neural networks

An artificial neural network (ANN) was developed to investigate whether hoof wall deformation could be used to determine ground reaction forces (GRF) in horses. The ANN was taught this relationship under certain conditions and was able to generalise this knowledge to conditions for which it was not trained before. To acquire data to train and test the ANN, a horse was equipped with strain gauges attached to the dorsal, lateral and medial parts of the hoof to assess hoof wall deformation. A force plate was used to measure the GRFs. Both hoof wall deformation and GRF were recorded simultaneously at different speeds, gaits, surfaces and loads. An ANN was trained with some of these data, and subsequently provided with strain gauge recordings of strides, not used for training. By comparing the GRF calculated by the ANN based on the hoof wall deformation with that recorded simultaneously by the force plate, the generalisability of the ANN was determined. It was found that an ANN is capable of 'learning' the relationship between hoof wall deformation and GRF, and to generalise it to a wide range of new conditions. This technique enables assessment of GRF under difficult conditions, such as on a treadmill or on surfaces where a force plate cannot be employed.