Abstract: In this paper，on the basis of data ACTG320 and 193A provided by ACTG，it is given that the study on curative effect prediction and therapy assessment of AIDS by using stochastic process theory and statistic analysis methods.It consists of first，giving the discrimination that data ACTG320 according to time series is a stochastic process obeying by normal distribution on cross face，discussing changeable relations of mean time series LCD4（t） and HIV（t） by quadratic fit and their correlation analysis which is highly significant，and getting the optimal stop-time model of a therapy；Second，on the basis of data 193A，investigating changeable relations of logarithmic type data（i.e.Ln（CD4+1）） means time series of four therapies with time t respectively and determining their optimal stop-time；finally，setting up the comprehensive assessment model of treatment affect and giving the comprehensive assessment of four therapies in data 193A.

Key words: stochastic process, correlation analysis, curve fitting, significant tests, optimal objective function

摘要： 在美国艾滋病医疗试验机构ACTG公布的两组数据ACTG320和193A的基础上，利用随机过程理论及数理统计分析方法进行了研究：首先给出了数据ACTG320按时间序列是一个其断面数据服从正态分布的随机过程的判断，通过数据拟合方法讨论了CD4浓度的对数型均值时间序列 LCD4（t）和HIV浓度均值时间序列HIV（t）的变化关系、相应的显著性检验及它们的相关性分析，给出了治疗的最佳终止时间的模型研究；其次在数据193A的基础上，进行了4种治疗方案的对数型CD4均值增量与时间t的关系变化研究，及相应治疗方案最佳终止时间的确定；最后建立了治疗效果的综合评价模型，就193A提供的4种治疗方案给出了综合评价。

关键词: 随机过程, 相关性分析, 数据拟合, 显著性检验, 优化目标函数