Modelling MRI enhancing lesion counts in multiple sclerosis using a negative binomial model: implications for clinical trials
Modelling MRI enhancing lesion counts in multiple sclerosis using a negative binomial model: implications for clinical trials
复制标题
DOI：
10.1016/s0022-510x(99)00015-5
复制
发表时间：
1999-02-01
影响因子：
4.4
通讯作者：
Filippi, L
中科院分区：
文献类型：
--
作者：
Sormani, MP;Bruzzi, P;Filippi, L
在多发性硬化症（MS）中，每月一次的磁共振成像（MRT）扫描中观察到的新增强病变的数量是实验性治疗的MRI监测研究中最广泛使用的反应变量。然而，尚未提出统计模型来描述MS患者中此类病变数量的分布。本文简要总结了计数数据的统计模型。提出负二项（NB）模型来拟合一组56例未治疗MS患者（随访9个月）中计数的新增强病变数量。结果表明，NE模型（残差偏差=66.6，54个自由度）比Poisson模型（残差偏差=1830.1，55个自由度）更好地解决了该数据集中存在的较大变异性。病变计数的参数化的应用程序进行了讨论，并提出了一个例子相关的计算机模拟的样本量估计。(C)1999 Elsevier Science B.V.保留所有权利。
In multiple sclerosis (MS) the number of new enhancing lesions seen on monthly magnetic resonance imaging (MRT) scans is the most widely used response variable in MRI-monitored studies of experimental treatments. However, no statistical model has been proposed to describe the distribution of the number of such lesions across MS patients. This article briefly summarizes the statistical models for counted data. The negative binomial (NB) model is proposed to fit the number of new enhancing lesions counted in a set of 56 untreated MS patients followed for 9 months, It is shown that the large variability present in this data set is better addressed by the NE model (residual deviance=66.6, 54 degrees of freedom):than by the Poisson model (residual deviance=1830.1, 55 degrees of freedom). Applications of the parametrization of lesion counts are discussed, and an example related to computer simulations for the sample size estimation is presented. (C) 1999 Elsevier Science B.V. All rights reserved.