加权Fisher线性判别法在非平衡医学数据集中的应用

目的:介绍加权Fisher线性判别法在非平衡医学数据集中的应用。方法:在两类分类问题中,当两类样本的协方差矩阵不同时,样本不平衡会导致Fisher线性判别的性能下降,使用加权Fisher线性判别法对两类样本同时进行不同倍数的过抽样,可促使两类的样本数目趋向平衡。结果:利用社区居民的血糖流行病学调查资料进行验证,加权Fisher线性判别法较传统Fisher线性判别法的灵敏度高,分类性能明显提高。结论:加权Fisher线性判别法可适用于非平衡数据集,算法简单高效,且基本不增加计算复杂度。     

线性回归方程的一种稳健求法

推荐一种常用于回归分析中的稳健估计方法 M估计 ,并结合实例用该方法对含有异常值的数据进行回归分析 ,克服了用传统的最小二乘法所产生的严重偏差 ,获得了较好的回归估计效果。     

重度妊高征实验室诊断方法研究

本文通过对重度妊高征患者血尿酸及平均动脉压的测定,依据多元分析的Fisher判别法,建立了一个定量分析的判别函数式,并给出了判别界值,根据孕妇的血尿酸及平均动脉压对重度妊高征进行定量的综合诊断。     

超声误诊卵巢囊肿的Fisher判别

目的构建超声误诊的卵巢囊肿性质的Fisher判别函数,提高诊断的敏感性和特异性。方法采用前瞻性研究方法,纳入132个超声误诊的卵巢囊肿,其中非赘生性囊肿53个,卵巢良性肿瘤47个,卵巢恶性肿瘤32个,以声像表现、血流信号、临床症状和妇科检查指标作为鉴别诊断变量,采用Fisher投影判别建立诊断模型。结果 3类卵巢囊肿的Fisher投影位置基本清晰;模型诊断非赘生性囊肿、良性肿瘤和恶性肿瘤的敏感性分别是78.9%、65.1%和71.9%(交叉性考核法)。结论 Fisher逐步判别分析为超声误诊的卵巢囊肿患者提供了较好的诊断模型,加强声像认识,详细询问病史,结合临床症状,采用模型判别等综合考虑是减少误诊的关键。     

回归直线的稳健求法及模拟研究

经典最小二乘法是求直线回归方程的常用方法,但当数据中存在异常值时,这种方法较敏感,本文给出了二类求稳健回归直线的方法,并将它们统一在迭代再加权最小二乘统一算法之下,模拟结果及实例分析表明这些方法在抗异常值方面较最小二乘法为优。     

基于多元素指纹的知母产地溯源技术研究

摘 要 目的：基于多元素指纹分析技术,对亳州和河北知母的产地进行判别研究,并建立判别模型。方法: 利用电感耦合等离子体质谱法（ICP MS）测定亳州和河北两个主产区44批知母样品中48种元素的含量,结合主成分分析（PCA）和正交偏最小二乘法判别分析（OPLS DA）,筛选出具有显著差异性的元素,对知母产地进行Fisher线性判别分析,并建立判别模型。结果:通过PCA、OPLS-DA筛选出9个差异性元素指标（Na、Mg、Rb、Zn、Sr、Mo、Sb、K、Cs）用于全模型法判别分析,建立了2个知母产地判别模型。所建立的判别模型经交叉验证,95.5%已分组观察值已正确地分类,回代验证,分组正确率为100%。结论:多元素指纹分析技术结合多元统计学方法是用于知母产地判别的一种有效方法,建立的判别模型良好,可应用于实际生产中知母产地的区分。     

心电图机几种干扰故障的排除方法

心电图机几种干扰故障的排除方法（山东省潍坊市坊子８９医院２６１２００）王全玉，李效银，李新立，郭力干扰，是电心图机最常发生且较难判别原因的故障。现将我们对几种干扰故障的判别及排除方法介绍如下。１．接插件接触不良干扰：因某一部分的接插件接触不良、排线内...     

多元回归分析肺吸虫囊蚴脱囊影响因素中有关异常点的探讨

目的:探讨准确发现、合理处理实验资料中异常点,资料满足多元线性回归分析条件,确保分析结果真实可靠。方法:采用残差图、残差分析、影响分析、稳健回归及根据专业知识判断。结果:第一组资料所对应的点为异常点,删去异常点,最小二乘法建立的方程与稳健回归建立的方程一致。结论:多种方法结合判断异常点,进一步提高了专业结论的科学性。     

临床常用药物对实验室检验结果的干扰

［摘要］论述药物对临床常规检验的干扰及药物本身代谢产物引起人体生理等方面的变化，出现假阳性或假阴性、假正常值或假异常值，或多种无法解释的结果，探讨影响因素，研究消除干扰的方法.     

儿童Miller Fisher综合征的临床分析

目的：分析儿童Miller Fisher综合征的临床特点，加深对本病的认识，避免误诊、误治。方法：回顾性分析近5年我院诊断为Miller Fisher综合征的6倒惠儿的临床特点，并结合文献进行讨论。结果：6倒Miller Fisher综合征患儿中眼肌麻痹5例，共济失调6倒，面瘫3例，球麻痹3例，腱反射消失5例；4例头颅CT／MRI正常，1例脊髓MPd提示神经根肿胀，神经传导异常4例，脑脊液有蛋白一细胞分离5例。结论：Miller Fisher综合征是格林．巴利综合征的变异型，以眼肌麻痹、共济失调、腱反射消失为特点。伴脑脊液蛋白增高和神经传导异常．该病预后多良好。     

A modification of the jump-flinch technique for measuring pain sensitivity in rats

The jump-flinch procedure provides a sensitive alternative to the hot-plate and tail-flick procedures. Analysis of the components of motor responses to increasing intensity of foot shock presentation has allowed the observational discrimination of five reliably elicited categories of unlearned responses to inescapable foot shock. Morphine sulfate differentially altered response category thresholds in rats. Response category thresholds also differed between Wistar and Fisher strain rats in analgesic effects of morphine sulfate.     

The novel-response procedure in humans.

The novel-response drug discrimination procedure is one of several three-choice procedures developed to address interpretational difficulties that can occur under standard two-response procedures. The novel-response procedure is unique among three-choice discrimination procedures by using instructions, rather than explicit training procedures. With the novel-response procedure, participants are trained under a standard two-response (drug vs. placebo) discrimination, and then instructed that in the presence of a drug stimulus unlike either of the training drugs, responses should be made on the novel-response alternative. Several studies have assessed the utility of the novel-response procedure by comparing effects under a standard two-response and the novel-response procedure in participants trained to discriminate triazolam from placebo. Results indicate that the novel-response procedure can increase the selectivity of both placebo- and drug-appropriate responding, and in this way, allows for finer distinctions to be made among sedatives than a standard two-response procedure.     

叶酸水分测定方法的研究

目的: 对叶酸的水分测定方法进行研究。方法:采用热重分析法(TG)、差示扫描量热法(DSC)等方法对叶酸的热力学行为进行研究,并建立了一个新的卡式炉-卡尔费休法测定水分的方法。结果: 采用新建立的方法测定国内及美国药典会(USP)叶酸对照品水分,结果分别为8.13%与8.00%,标准偏差分别为0.74%与0.28%(n=3)。结论: 本文建立的卡式炉-卡尔费休法方法简单方便,准确可靠,适合批量样品的测定,为叶酸对照品的标定及其原料的质量控制提供依据。     

The assessment of the action of pharmacological agents based on a new computer program for analysing animal operant behavior

The new programme is developed for the experimental automatization and analysis of animal operant behavior in Skinner's box (Lafayette Instrument Co., USA) by means of the Apple 2e computer (USA). The fundamental of the programme is the division of the training procedure into different functional intervals. Operant behavior of rats is determined by diverse schedules of food and water reinforcement and electric shock. Rats were trained to response on schedules FR 20, FI 1, drug discrimination. Phenazepam (2 mg/kg) markedly decreases the number of responses on schedule FR 20. Phenazepam is a discriminable stimulus. The experimental results make it possible to use the programme for modelling various forms of operant behavior and analysing pharmacological properties of the well-known and new drugs.     

Cocaine-like discriminative stimulus effects of procaine,dimethocaine and lidocaine in rats

The discriminative stimulus effects of several local anesthetics and (+)-amphetamine were assessed in a drug discrimination based on the psychomotor stimulant cocaine. Two groups of adult male Sprague-Dawley rats were trained in two-lever operant chambers in a cocaine versus saline discrimination, or a cocaine versus procaine or saline discrimination, using a fixed ratio 20 schedule of food reinforcement. Cocaine, (+)-amphetamine and dimethocaine all dose-dependently substituted for the training dose of cocaine (10 mg/kg) in both procedures. While procaine and lidocaine showed partial substitution in the cocaine versus saline procedure, much less substitution occurred in the cocaine versus procaine or saline discrimination. These data demonstrate that it is possible to train rats to discriminate between cocaine and another local anesthetic procaine, resulting in an increased pharmacological specificity of cocaine discrimination. The fact that dimethocaine fully substituted for cocaine in both procedures indicates that this local anesthetic has more cocaine-like effects than others so far tested, which is consistent with results from other behavioral and neurochemical studies with this compound.     

计算机实时监控大鼠药物辨别实验

本文工作利用计算机及有关功能板与辨别实验箱构成了一个《计算机实时监控大鼠药物辨别实验》装置,所编制的计算机程序实现了1至6个辨别实验箱实验前系统检测、实验的启动和终止、实验中实时控制和监测以及数据的计算、整理和存储打印的自动化。在将大鼠装入实验箱后,实验者只需坐在计算机前,通过键盘在计算机的详尽提示下键入几个所选定项目号和提供几个必需的数据,实验就能够自动完成。这些,使本系统不但易于一般药理工作者使用,还大大减少了实验者的工作量;由于在很大程度上避免了人为介入,因此降低了产生偶然误差的机会,提高了大鼠药物辨别实验的客观性。     

Titration of duration discrimination in behavioral pharmacology and toxicology

This study investigated a discrete-trial, titration duration discrimination procedure in behavioral pharmacology. Pentobarbital and d-amphetamine, measured with this procedure, selectively affected discrimination more than response tendencies. Pentobarbital also tended to affect selectively discrimination of longer durations, whereas d-amphetamine did not. Further experiments showed that (1) other algorithms for modulating stimulus duration are useful in behavioral pharmacology and toxicology, (2) threshold estimates are similar with the method of constant stimuli and the method of titration, and (3) this titration procedure permits the separate examination of drug effects upon discrimination and upon response tendencies; the fixed-interval procedure does not. Baseline variability was an important correlate of drug effects in that the endpoints with more variable baselines were also more sensitive to drugs.     

Drug-induced state-dependent learning: review of an operant procedure in rats

Drug discrimination and drug state dependence are often thought to be based on the same drug actions, and to differ only in the doses needed to produce them, with discrimination occurring at low doses and state dependence at high doses. Testing this hypothesis has been hampered by the use of discrimination and state dependence procedures that differed in many respects. In 1986, Colpaert introduced a procedure to study state dependence in rats that used the same response, the same reinforcer, and the same reinforcement schedule that are commonly used in drug discrimination. Using this procedure, differences between drug state dependence and drug discrimination were found with some drugs (e.g. alcohol), consistent with the hypothesis that the procedures differ in the drug properties they measure, but not with other drugs (e.g. chlordiazepoxide). Thus, state dependence and drug discrimination can generate different outcomes, but the conditions in which they do require further study. However, all the studies conducted with the procedure introduced by Colpaert clearly show that state dependence is not necessarily only a high-dose phenomenon, but can also occur at doses at which many central nervous system drugs produce their characteristic effects. This finding led to the hypothesis that state dependence may be involved in the therapeutic and other effects of psychoactive drugs.