Body thiamine level in acute alcohol psychoses

PURPOSE: To compare admission data and academic performances of medical students younger and older than 25, and to qualify older students' experiences and perceptions in medical school. METHOD: The authors reviewed 1988-1991 data for applications to the McGill University Faculty of Medicine. Data included GPAs and MCAT scores, as well as ratings for reference letters, autobiographical statements, and interviews. For those same years, the authors measured students' academic performances in the preclinical and clinical years. The authors compared the data by students' age: "younger" students, aged 17 to 24; and "older" students, aged 25 and above. All enrolled students took the Derogatis Stress Profile, and the older students participated in focus groups. RESULTS: The older applicants had lower GPAs and MCAT scores, but higher interview and reference letter ratings. For older accepted students, basic science course scores were lower than those of younger students, but clinical scores did not differ significantly between the groups. The two groups had similar stress levels, although older students tested lower in driven behavior, relaxation potential, attitude posture, and hostility. In focus groups, the older students spoke of learning style differences, loss of social support, and loss of professional identity. CONCLUSION: Different scores in admission criteria suggest that McGill uses different standards to select older medical students. Older students admitted under different criteria, however, do just as well as do younger students by their clinical years. A broad-based study of admission criteria and outcomes for the older student population is warranted.     

神经网络极速学习方法研究进展

神经网络已经在模式识别、自动控制及数据挖掘等领域取得了广泛的应用,但学习方法的速度不能满足实际需求。传统的误差反向传播方法(BP)主要是基于梯度下降的思想,需要多次迭代;网络的所有参数都需要在训练过程中迭代确定,因此算法的计算量和搜索空间很大。ELM(Extreme Learning Machine,ELM)是一次学习思想使得学习速度提高很多,避免了多次迭代和局部最小值,具有良好的泛化性能、鲁棒性与可控性。但对于不同的数据集和不同的应用领域,无论ELM是用于数据分类或是回归,ELM算法本身还是存在问题,所以本文对已有方法深入对比分析,并指出极速学习方法未来的发展方向。     

磁能驱动微型泵的性能研究

在电磁驱动原理的基础上,设计并研制了一种磁能驱动的微型泵。微型泵的整体尺寸约为Φ11mm×4mm,腔室半径为5mm,深2mm。利用方差分析方法对实验数据进行了检验,验证了前期微型泵研究中获得的性能参数的正确性。为了进一步缩小微型泵体积、简化控制方式和外围电路,针对微型泵的能源部分,开展了单双电源的对比实验研究,并获得了较好的实验结果。     

基于CAN总线的智能远程温度开关设计

针对多点温度远程控制问题,设计具有CAN总线通信、键盘输入和数显功能的智能远程温度开关,介绍了温度测量的基本原理,给出相应的硬件电路与单片机软件实现.实验结果表明:精度达3.9×10-3,响应时间低于6s,该温度开关误差小、性能稳定,能够满足整体性能要求.     

基于Adaboost-CSHG的特定类目标跟踪识别

目标跟踪识别是计算机视觉领域的热点研究对象。首先采用基于Adaboost的目标检测算法,训练得到了特定类目标坦克模型的级联分类器,对图像中的坦克目标完成了 “粗检测”;通过构建类属超图(CSHG)模型,采取Adaboost与CSHG相结合的方式,有效滤除了大量虚警,实现了对坦克目标的“精检测”,同时完成了对目标的跟踪;最后利用基于类属超图的目标识别原理对目标进行识别,实验结果表明该方法在简单背景和复杂背景图像条件下均具有可行性。     

一种多约束的密度聚类算法的研究

针对传统的密度聚类算法不能处理带有多约束条件的问题,在现有的密度聚类算法的基础上,提出了一个带有多约束条件限制的密度聚类算法。该算法将多约束条件引入到密度聚类分析中,并分析了多约束条件对聚类结果的影响。实验表明该算法在多约束条件下,可有效完成对数据点的聚类并且效果较好,为现实情况中处理多约束聚类提供了良好的理论支持。     

一种网络多点备份系统的设计和实现

该文提出并实现了一种网络多点备份系统。通过在本地应用服务器的设备驱动层之上加入写监控驱动,所有写操作将被封装成备份记录并发往本地缓存网关暂存,然后由本地缓存网关异步地将备份记录发送到多个处于不同地域的远程备份网关,最后在远程备份网关上重建写操作。该系统实现了对本地数据的多点异地备份,能够适应多种网络状况,支持Linux下所有文件系统和物理存储设备     

Infections due to Lancefield group F and related Streptococci (S. milleri, S. anginosus)

We can no longer accept classification of streptococci solely on the basis of hemolytic reactions or Lancefield agglutinations. While the "viridans" streptococci cannot be serologically differentiated, physiological separation of the species offers a satisfactory method of classifying human isolates. We review the microbiology of Lancefield group F and related streptococci (S. milleri, S. anginosus), emphasizing microbial ecology and current taxonomic considerations. A series of 28 patients infected with these organisms is presented. There was a striking male predominance in the series (6:1) as well as an obvious association with underlying diseases and/or antecedent trauma. The most remarkable feature of these pathogens is their predilection for abscess formation, confirming their overdue recognition as the most common "viridans" streptococcus recovered from abscesses within internal organs. We observed purulent disease of the nervous and skeletal systems, oral cavity, lung and pleural space, abdomen and subcutaneous tissues. Microbial synergy (i.e. polymicrobic infection) was not a requisite for abscess formation. Four of the five deaths in this series occurred in patients 60 years of age of older. Some degree of variability in antimicrobial susceptibility was noted, so speciation alone may not always provide sufficient information on which to base a therapeutic decision.     

双端输出塑料闪烁体探测器的符合关联测试

利用双端信号输出的塑料闪烁体与灵敏度高、时间响应快的硅光电倍增管构成塑料闪烁体探测器探头,并与后端的数字化转换器等电子学系统搭建成塑料闪烁体探测器系统。为了研究不同数据处理方法对真事件探测效率和能量分辨率的优化,分别使用标准γ源60Co和137Cs对塑料闪烁体探测器系统进行测试。研究了关联事件的符合时间窗对探测器真事件探测效率的影响,分析了波形的积分长度与脉冲信号甄别(Pulse Shape Discrimination,PSD)开窗法对能谱能量分辨率的改进。结果表明,在关联事件的符合时间窗为15 ns时,真事件探测效率最佳,当波形积分长度为80 ns时,通过PSD开窗后能量分辨率由原来的53.38%优化为42.21%。     

群机器人仿真系统设计与实现

群机器人仿真系统是一个能够对群机器人系统的体系结构、协调控制和学习算法等进行实验的数字仿真系统。分析了各仿真对象的物理特性及其实现难点,给出了仿真系统的软件架构和对象建模,描述了各仿真功能的具体实现。利用本仿真系统进行的实验结果表明此系统达到了比较理想的效果,具有良好的分布式运行能力和方便的实时交互能力,为群机器人系统的研究提供了一个有效的平台。