基于线性混合模型的艾滋病最佳治疗时机选择

Linear Mixed Effect Model for the Best AIDS Treatment Timing

本文利用美国艾滋病医疗试验机构ACTG的193A研究中的一组非平衡重复测量数据。以log(cd4+1)为体现疗效的因变量,年龄、性别为固定效应,治疗时间和滞后治疗时间为随机效应,同时考虑疗法对疗效的影响引入其与治疗时间的交互效应,建立线性混合效应模型。用SAS软件求解。再通过建立以治疗时间斜率随机效应为因变量初始logcd4为解释变量的回归模型判断艾滋病最佳治疗时机。结果表明,当初始cd4为185个/mm^3时治疗时机最佳,即为无症状感染的晚期.与美国DHHS推荐的小于200个/mm^3一致,却更为科学和精确。本研究对艾滋病治疗的临床实践具有重要的指导意义。

In this paper,we use a group of non-equilibrium repeated measurement data from the 193A study of ACTG,logcd4 to reflect the curative effect of the dependent variable.Taking age,gender as fixed effect treatment time and lag time as random effect,for taking into account the impact of therapy on the efficacy and introducing its interaction with the treatment time,we establish linear mixed model. We use SAS to estimate.Through the establishment of regression model that the slope of treatment time as dependent variable initial logcd4 as explanation variable,we determine the best AIDS treatment time. The results show that When the initial 184.7317 for cd4,the best time for treatment,which is the later stage of none symptom infection.This is less than recommended by the U.S.DHHS 200/mm^3 consistent, but more scientific and accurate.This study has an important significance for the clinical practice of HIV treatment.