Nonlinearity identified by neural network models inP CO 2 control system in humans

The nonlinearity included in the P_{CO 2} control system in humans is evaluated using the degree of nonlinearity based on a difference of residuals. An autoregressive moving average (ARMA) model and neural networks (linear and nonlinear) are employed to model the system, and three types of network (Jordan, Elman and fully interconnected) are compared. As the Jordan-type linear network cannot approximate respiratory data accurately, the other two types and the ARMA model are used for the evaluation of the nonlinearity. The results of the evaluation indicate that the linear assumption for the P_{CO 2} control system is invalid for three subjects out of seven. In particular, strong nonlinearity was observed for two subjects.     

Standard skin prick testing and sensitization to inhalant allergens across Europe--a survey from the GALEN network

Skin prick testing (SPT) is the standard method for diagnosing allergic sensitization but is to some extent performed differently in clinical centres across Europe. There would be advantages in harmonizing the standard panels of allergens used in different European countries, both for clinical purposes and for research, especially with increasing mobility within Europe and current trends in botany and agriculture. As well as improving diagnostic accuracy, this would allow better comparison of research findings in European allergy centres. We have compared the different SPT procedures operating in 29 allergy centres within the Global Allergy and Asthma European Network (GA(2)LEN). Standard SPT is performed similarly in all centres, e.g. using commercial extracts, evaluation after 15-20 min exposure with positive results defined as a wheal >3 mm diameter. The perennial allergens included in the standard SPT panel of inhalant allergens are largely similar (e.g. cat: pricked in all centres; dog: 26 of 29 centres and Dermatophagoides pteronyssinus: 28 of 29 centres) but the choice of pollen allergens vary considerably, reflecting different exposure and sensitization rates for regional inhalant allergens. This overview may serve as reference for the practising doctor and suggests a GA(2)LEN Pan-European core SPT panel.     

基于主成分分析-神经网络的非编码RNA预测

预测非编码RNA对认识其调控功能具有重要意义。选择单核苷酸和二核苷酸出现频率作为神经网络分类特征,运用主成分分析方法降低输入数据的维数,去除数据间的相关性,采用Levenberg-Marquardt算法改善网络训练速度。对数据集的测试结果表明,此方法对细菌混合ncRNA的预测精度达到81.3%,对原核生物tRNA的预测精度达到93.3%,表明该方法能有效预测原核生物ncRNA。预测结果还发现两种古细菌的ORF序列在序列特征上与其它细菌和古细菌存在差别。     

舌粘膜及粘膜下血管网的构筑及其临床意义

本研究应用手术显微镜和扫描电镜观察了人舌粘膜的微血管构筑。舌深动脉长支和舌背动脉终支在舌粘膜下形成一完整的致密动脉网,跨越界沟和舌正中线,成为一整体。由动脉网发出微动脉支进入舌粘膜内,形成粘膜内的毛细血管网和各种乳头毛细血管丛。另外,还讨论了舌粘膜下馓血管构筑和舌瓣设计与临床舌诊的关系。     

基于多分辨率分析的心室晚电位神经网络检测

采用多分辨率分析和人工神经网络相结合的方法实现对心室晚电位的检测。首先利用多分辨率分解技术提取高分辨率心电信号不同频带的能量构成一组特征值 ,再利用这些特征值训练BP神经网络 ,并完成对心室晚电位的识别。经过对 2 8例 3导高分辨率心电图实验数据的处理 ,取得了较高的识别准确率     

化合物构效关系的模糊极小极大神经网络识别研究

应用模糊极小极大神经网络研究了化合物复杂结构和性能（QSAR）之间的关系，用该法进行几组化合物致癌的识别，结果优于线性回归的方法，对此作出一些分析。     

一种EEG信号盲分离和分类的神经网络方法

提出一种采用多神经网络处理脑电(EEG)信号的方法。首先，对混有噪声的脑电信号给出一种盲分离的自适应神经算法。通过寻求采样时间序列线性组合的kurtosis系数的局部极值，得出该算法的模型和步骤。在盲分离的基础上，对分离出的估计信号进一步利用Kohonen网络进行分类。将该算法用于300个EEG样本处理，并给出处理结果。     

抗聚合人血清白蛋白的抗独特型抗体

一种针对兔抗聚合人血清白蛋白抗体(anti-PHSA,Ab_1)的抗独特型抗体(anti-anti-PHSA,Ab_2),已通过免疫同种动物制成。该Ab_2具有以下特性:(1)它既可与兔anti-PHSA反应,也可与鼠的单克隆的anti-PHSA反应。(2)它与鼠的单克隆的anti-PHSA的反应,可受到PHSA和乙型肝炎病毒(HBV)上的聚合人血清白蛋白受体(PHSA-R)的抑制。(3)它可在异种动物体内诱导产生具有anti-PHSA活性的抗-抗独特型抗体(anti-anti-anti-PHSA,Ab_3),而且在收获的Ab_2血清中也查见了Ab_3。本研究不仅表明了针对PHSA的免疫应答中独特型网络调节的客观存在,而且在其抗独特型抗体(anti-Id)中,的确存在具有PHSA内映像的成分。     

国人延髓动脉的观察

本文观察了110例成人脑标本的延髓外部动脉;采用动脉X线造影法和透明法,在20例人脑标本上观察了延髓内部动脉的形态和供应。将延髓表面分为四区,供应各区的外部动脉可分为相应的4群。前正中群和前外侧群动脉主要来自脊髓前动脉,前正中群动脉供应舌下神经核、内侧丘系及部分锥体束等中缝两侧结构;前外侧群动脉供应大部分锥体束;外侧群动脉来自椎动脉、小脑下后动脉、基底动脉和小脑下前动脉,主要供应网状结构、脊髓丘脑束、三叉神经脊束及核、迷走神经背核等;后群动脉来自脊髓后动脉和小脑下后动脉,主要供应薄、楔束核等。本文还讨论了延髓外侧区的血液供应及“终动脉”等问题。     

HOPFIELD网络的联想记忆性质

本文首先给出了有关联想记忆的一系列概念，通过计算机模拟验证了网络的某些固有性质，同时指出了它的固有问题，并仿真研究了Ｈｏｐｆｉｅｌｄ网络中参数间的关系。