Pharyngeal wall vibration detection using an artificial neural network

OBJECTIVES: To assess the likelihood of health care legislation in the forthcoming 105th Congress in 5 areas: health care coverage, tax and Employee Retirement Income Security Act (ERISA) policy, Medicaid, Medicare, and managed care. DESIGN: Informal, semistructured conversations that took place in the months prior to the 1996 elections. POPULATION: Congressional health staff and administration officials. OUTCOME MEASURES: Unofficial, off-the-record personal opinions. RESULTS: Health care coverage initiatives to benefit children and unemployed persons are likely to be proposed, but have little chance of enactment; children are seen as well provided for under current Medicaid law, the strong economy and high employment levels lower concern for unemployed issues, and the effort required to pass the Kassebaum-Kennedy legislation needs time to settle. Tax proposals, such as medical savings accounts (MSAs), and ERISA amendments have no constituency; also, the MSA demonstration in Kassebaum-Kennedy will forestall further action. Medicaid is far less an issue than in the previous Congress, because spending has fallen unexpectedly, the bitter fight over block grants makes them unlikely to be revisited, and the administration is likely to enhance state flexibility through waivers. Medicare will be the subject of substantial action to defer impending insolvency temporarily, but there is virtually no chance that definitive long-term solutions will be enacted even though the underlying fiscal problems are thoroughly understood and recognized. Managed care will be the venue for numerous proposals designed to address specific consumer and quality issues. CONCLUSIONS: Four bitter years of fighting over health care issues has raised awareness of the problems, but has produced a political chemistry that is too rancorous to permit passage of significant legislation in the near future.     

Quantification of optic nerve blood flow changes using magnetic resonance imaging

PURPOSE: To evaluate the possibilities of magnetic resonance imaging (MRI) for quantification of pharmacologically induced changes in optic nerve microcirculation. METHOD: T2-weighted MRI sequences were used to image the eye, optic nerve, and frontal cortex in rats. Two sets of control images before and one set during Gd (DTPA) infusion were recorded. Blood flow values for two regions of the optic nerve (an anterior part, including the optic nerve head, and a more posterior part) and the frontal cortex were calculated by image analysis from the change in signal intensity, as already reported for cerebral blood flow. For each rat, a control experiment before drug administration and a second experiment 30 minutes after subcutaneous injection of either placebo (n = 7), timolol (n = 7), or SDZ GLC-756, a dopamine D-1 antagonist and D-2 agonist (n = 7), were carried out in a double-blind fashion. RESULTS: Mean basal blood flow values were found between 29.4 and 45.6 ml/100 g per minute in the anterior part of the optic nerve, 38.3 and 42.9 ml/100 g per minute in the posterior part of the optic nerve, and 68.0 and 75.0 ml/100 g per minute in the frontal cortex. Placebo and timolol did not cause significant changes. SDZ GLC-756 significantly increased blood flow by 238% +/- 65% in the anterior part and by 87% +/- 40% in the posterior part of the optic nerve. CONCLUSIONS: These results suggest that MRI provides quantification of optic nerve blood flow and that dopaminergic substances increase optic nerve blood flow.     

基于BP神经网络的供应链突发风险评价模型

为评价供应链突发风险的影响,从供应链的组成和运作要素角度定义了5种类型的供应链突发风险,并建立了一套基于风险类型的供应链突发风险评价指标体系.在此基础上,以风险评价指标为输入,风险类型和风险等级为输出,建立了基于BP神经网络的供应链突发风险评价模型.该模型使用BP网络作为经验知识的学习机制来学习突发风险评价指标与突发风险类型及等级之间的映射关系,然后通过BP网络的知识记忆和泛化推理能力实现对供应链突发风险的自动识别和评价.提出的模型在一个供应链风险样本集上进行了验证,验证结果表明了模型的有效性.     

A neural network based artificial vision system for licence plate recognition

This paper presents a neural network based artificial vision system able to analyze the image of a car given by a camera, locate the registration plate and recognize the registration number of the car. The paper describes in detail various practical problems encountered in implementing this particular application and the solutions used to solve them. The main features of the system presented are: controlled stability-plasticity behavior, controlled reliability threshold, both off-line and on-line learning, self assessment of the output reliability and high reliability based on high level multiple feedback. The system has been designed using a modular approach. Sub-modules can be upgraded and/or substituted independently, thus making the system potentially suitable in a large variety of vision applications. The OCR engine was designed as an interchangeable plug-in module. This allows the user to choose an OCR engine which is suited to the particular application and to upgrade it easily in the future. At present, there are several versions of this OCR engine. One of them is based on a fully connected feedforward artificial neural network with sigmoidal activation functions. This network can be trained with various training algorithms such as error backpropagation. An alternative OCR engine is based on the constraint based decomposition (CBD) training architecture. The system has showed the following performances (on average) on real-world data: successful plate location and segmentation about 99%, successful character recognition about 98% and successful recognition of complete registration plates about 80%.     

Latent inhibition: A neural network approach.

A formal theory of latent inhibition (LI) is offered in the context of a real-time, neural network model of classical conditioning. The network assumes that the effectiveness of a CS in establishing associations with the unconditioned stimulus/stimuli (UCS) is proportional to total novelty, defined as the sum of the absolute value of the difference between the predicted and observed amplitudes of all environmental events. CS effectiveness controls both the rate of storage (formation, or read-in) and the retrieval (activation, or read-out) of CS-CS and CS-UCS associations. The model describes LI because total novelty and, therefore, CS effectiveness decrease during CS preexposure. Computer simulations demonstrate that the neural network correctly describes, and sometimes predicts, the effects on LI of experimental manipulations before and during CS preexposure and during and after conditioning. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

A neural network model of the McCollough effect

Recent federal legislation has contained the stipulation that participating health maintenance organizations (HMOs) include a quality assurance program which stresses health outcomes. This provision was ostensibly directed at correcting alleged abuses in HMOs serving the urban poor. One version of the outcome method was employed for an 18-month period at an urban HMO caring for 2,000 Medicaid subscribers. The program involved comparing diagnostic accuracy and therapeutic outcomes for clinical conditions relevant to the study population with ideal standards established by the HMO. Three conditions were selected: contraception, depression, and hypertension. The results revealed widespread underdiagnosis (44-74%) in each condition and unacceptable therapeutic outcomes in two. Data collection was hampered by shifts in geography and financial eligibility among the denominator population and low response rates (38-63%) to telephone and mail surveys. Applying the general project guidelines to specific conditions proved considerably more difficult than anticipated. Further refinement of this approach to quality assessment must occur before its widespread use is feasible. Its effectiveness in improving quality remains to be seen. This experience raises doubts regarding the wisdom of legislating a specific outcome approach to quality assessment before feasibility and effectiveness have been demonstrated in organized health settings.     

Discrimination of forearm''s motions by surface EMG signals using neural network

The effect of the antiviral, antitumoural xanthate D609 on the activity of phospholipase A2, C (PC- and Pi-specific) and D was investigated. D609 is the first model substance of a new concept of antiviral therapy that interferes with cellular regulation mechanisms, rather than with virus coded enzymes. Exclusively phosphatidylcholine (PC) specific phospholipase C (PC-PLC) was found to be inhibited in a dose-dependent manner. Enzyme activity was determined either as the rate of acid release from PC or as the rate of phosphorylcholine production form 3H labelled PC. Lineweaver-Burk plots revealed D609 as a competitive inhibitor of PC-PLC with a Ki of 6.4 microM. In addition, D609 competitively inhibited PC-PLC mediated cleavage of P-nitrophenylphosphorylcholine (p-NPP), a pseudo-substrate of PC-PLC with a Ki of 8.8 microM. These data suggest that D609 competes with the phosphorylcholine residue of PC for binding to PC-PLC.     

Large receptive fields for optic flow detection in humans

We used a psychophysical summation technique to study the properties of detectors tuned to radial, circular and translational motion, and to determine the spatial extent of their receptive fields. Signal-to-noise motion thresholds were measured for patterns curtailed spatially in various ways. Sensitivity for radial, circular and translational motion increased with stimulus area at a rate predicted by an ideal integrator. When sectors of noise were added to the stimulus, sensitivity decreased at a rate consistent with an ideal integrator. Summation was tested for large annular stimuli, and shown to hold up to 70 degrees in some cases, suggesting very large receptive fields for this type of motion (consistent with the physiology of neurones in the dorsal region of the medial superior temporal area (MSTd)). This is a far greater area than observed for summation of contrast sensitivity to gratings (Anderson SJ and Burr DC, Vis Res 1987;29:621-635, and to this type of stimuli (Morrone MC, Burr DC and Vaina LM, Nature 1995;376:507-509, consistent with the suggestion that the two techniques examine different levels of motion analysis.     

A neural network that predicts psychiatric length of stay

One hundred and twelve bilateral thyroidectomies for solitary thyroid nodules with suspected malignancy were performed. The incidence of malignancy in the 112 primary nodules was 42%. Twenty-nine of the 112 contralateral lobes (26%) contained malignancy, which was unsuspected in 80%. Of these 29, 20 were foci of papillary cancer < or = 5 mm and 9 were larger papillary tumors or follicular carcinoma. We analyzed these 112 patients to determine whether there was a simple method to identify those patients at risk for contralateral, unsuspected malignancy. With use of the AMES clinical staging retrospectively, 70 of the 112 patients were classified as having low-stage disease. Fifty percent (35) had cancers on the primary side and 27% (19) on the contralateral side. Of these 19 contralateral cancers, 14 were papillary cancers < or = 5 mm, 4 were papillary cancer > 5 mm and one was a 1.5 cm follicular carcinoma, a similar distribution as in the whole group of 112. When the AMES analysis then excluded those thought to be at risk for multicentricity or papillary carcinoma and examined female patients only with nonpapillary frozen sections, nonpapillary aspiration cytological results, and no history of radiation exposure, no further reduction in the proportion of contralateral cancers (7 of 26, 27%) was found. Fifty-five of the 112 patients underwent preoperative ultrasound scans. In those cases in whom the contralateral lobe had no intraoperative palpable or preoperative sonographic mass, 5 of 20 still had contralateral cancers, but all were papillary < or = 5 mm.(ABSTRACT TRUNCATED AT 250 WORDS)     

Bridging GPS outages of tightly-coupled GPS/SINS using GMDH neural network

A tightly-coupled GPS (global positioning system)/SINS (strap-down inertial navigation system) based on a GMDH (group method of data handling) neural network was presented to solve the problem of degraded accuracy for less than four visible GPS satellites with poor signal quality.Positions and velocities of the satellites were predicted by a GMDH neural network,and the pseudo-ranges and pseudo-range rates received by the GPS receiver were simulated to ensure the regular operation of the GPS/SINS Kalman filter during outages.In the mathematical simulation a tightly-coupled navigation system with a proposed approach has better navigation accuracy during GPS outages,and the anti-jamming ability is strengthened for the tightly-coupled navigation system.     

Application of fuzzy sets and neural network theory to the numerical pattern recognition of bubble flow and jet flow

The injection of gas into a liquid is associated with two characteristic flow regimes: bubble flow and jet flow. In metallurgy, these flow regimes are of interest with respect to refractory lifetime and mass transfer. The flow regimes were investigated in the systems water-air and ethanol-air by monitoring the bottom vibrations of a cold model. The vibrations were analyzed by numerical pattern recognition. The nominal Mach number of 1.1 is the criterion for the transition from bubble flow to jet flow. A numerical pattern recognition system which is able to distinguish between the flow regime below and above the nominal Mach number of 1.1, just by analyzing the vibration signal, is presented. This task can be accomplished in spite of the use of different liquids, nozzle diameters and liquid heights. As a classificator, both a fuzzy k nearest neighbour algorithm and a neural network were employed successfully. Fuzzy logic was found to be especially useful for describing the different flow regimes.     

神经网络控制系统的研究概况

本文介绍了神经网络控制系统的类型和研究概况，并对其存在的问题和发展趋势作了讨论。     

A constrained neural network Kalman filter for price estimation in high frequency financial data

In this paper we present a neural network extended Kalman filter for modeling noisy financial time series. The neural network is employed to estimate the nonlinear dynamics of the extended Kalman filter. Conditions for the neural network weight matrix are provided to guarantee the stability of the filter. The extended Kalman filter presented is designed to filter three types of noise commonly observed in financial data: process noise, measurement noise, and arrival noise. The erratic arrival of data (arrival noise) results in the neural network predictions being iterated into the future. Constraining the neural network to have a fixed point at the origin produces better iterated predictions and more stable results. The performance of constrained and unconstrained neural networks within the extended Kalman filter is demonstrated on "Quote" tick data from the $/DM exchange rate (1993-1995).     

模糊神经网络炉温自学习控制系统

提出一种模糊神经网络自学习控制方法，并应用于熟料窑炉温度控制系统中。经实验仿真和应用结果表明，该控制方案可改善具有时变及大纯滞后的炉温控制系统，其性能优于一般Ｆｕｚｚｙ控制。