非线性预测方法在艾滋病疗法效果研究中的应用

发布时间：2018-04-11 17:27

本文选题：非线性预测 + 神经网络　；参考：《成都理工大学》2007年硕士论文

【摘要】： 本论文针对艾滋病疗法效果的所得到的临床数据(ACTG320)，采用了定量预测方法对数据进行处理，寻找其内在规律性，拟得出艾滋病治疗方法对病人病情的影响，预测继续治疗的效果。论文先介绍了论题的相关背景知识，包括常见的艾滋病疗法及其效果，知道CD4数目和HIV浓度是衡量艾滋病患者病情的两个重要指标量，引出了本论文的主要任务就是用一些数学预测方法来模拟CD4和HIV之间非线性的关系。为了实现对艾滋病疗法效果的预测，论文以CD4数目初值、HIV浓度的初值和治疗时间段为输入项，CD4数目改变量、HIV浓度改变量为输出项，建立了BP神经网络。通过对原始数据进行分组和预处理，对网络进行训练和泛化能力研究，得到了合理的网络模型，将其应用于艾滋病疗法的长期治疗效果预测。针对不同的病人，预测了其继续治疗的效果，对于效果不好的，给出了提前终止治疗的时间。其次，论文分别以CD4数目初值、HIV浓度的初值和治疗时间段为输入项，CD4数目改变量或HIV浓度改变量为输出项建立了回归支持向量机模型。通过对样本的学习训练，将训练好的回归支持向量机应用于艾滋病疗效的预测。文中通过BP神经网络和回归支持向量机两种预测方法在艾滋病疗法效果研究上的应用，，经过对比，发现基于支持向量机的艾滋病疗法效果预测模型在估计精度、预测能力等方面都优于BP神经网络。
[Abstract]:According to the clinical data of AIDS therapy, the quantitative prediction method is used to process the data, to find out the inherent regularity, to obtain the effect of AIDS treatment on the patient's condition, and to predict the effect of continuous treatment.The paper first introduces the relevant background knowledge of the topic, including the common AIDS treatment and its effect. We know that the number of CD4 and the concentration of HIV are two important indicators to measure the condition of AIDS patients.The main task of this paper is to simulate the nonlinear relationship between CD4 and HIV by some mathematical prediction methods.In order to predict the effect of AIDS therapy, a BP neural network was established by using the initial value of CD4 number and the initial value of CD4 concentration and the time period of treatment as input items, CD4 number change quantity and HIV concentration change quantity as output item.Through grouping and preprocessing the raw data and studying the network training and generalization ability, a reasonable network model is obtained, which is applied to predict the long-term therapeutic effect of AIDS therapy.For different patients, the effect of continuous treatment was predicted, and the time of early termination of treatment was given.Secondly, the regression support vector machine model was established using the initial value of CD4 number and the initial value of CD4 concentration as the input item, CD4 number change or HIV concentration change as the output item, respectively.The trained regression support vector machine (RSVM) was applied to predict the effect of AIDS.Through the application of BP neural network and regression support vector machine (RSVM) in the study of the effect of AIDS therapy, it is found that the prediction model based on support vector machine (SVM) is accurate in estimating the effect of AIDS therapy.The prediction ability is better than BP neural network.
【学位授予单位】：成都理工大学
【学位级别】：硕士
【学位授予年份】：2007
【分类号】：R512.91;R311

【引证文献】