在医学领域研究中药对人体过敏准确判断问题。由于,中药对人体过敏的影响是一个较为漫长的过程,不同中药对不同人体的过敏性影响也不同。针对中药过敏信息存在各种干扰,关于建立不同类型的过敏数据联系,剔除无关数据干扰,传统建模很难。提出利用自回归支持向量机的人体对中药潜在过敏性估计方法。根据过敏数据主成份分析相关理论,将人体对中药过敏的初始变量进行标准化处理,计算相应的协方差矩阵,删除所有的可能的过敏判断干扰冗余数据,使得到的结果作为大数据分析的样本数据。用自回归模型和支持向量基模型进行融合,建立自回归支持向量基模型,再把过敏样本数据输入到模型中,实现人体中药潜在过敏性估计。实验结果表明,利用改进算法进行人体对中药潜在过敏性分析,能够提高分析的准确性。
An estimation method for the potential allergenicity of human body to traditional Chinese medicine is proposed based on the regression support vector machine (SVM). According to the related theory of principal compo- nent analysis for allergy data, the initial variables of the human body, which is allergic to traditional Chinese medi- cine, is standardized to calculate the corresponding covariance matrix and all possible redundant data interfered with allergic judgment are deleted. The obtained results are analyzed by an auto - regression support vector machine. The experimental results show that the improved algorithm can increase the accuracy of the analysis.