摘要
针对中药多维度、小样本、非线性作用机制及成分间复杂关系难以解析的问题,文章提出融合网络特征的偏最小二乘分析方法。通过构建中药特征网络,采用Louvain社区识别成分协同-拮抗关系,结合Moreau-Yosida正则化与Nesterov优化实现特征筛选,建立改进的偏最小二乘回归模型提取非线性成分。在8组UCI数据集和麻杏石甘汤止咳、平喘、退热实验中进行验证,结果表明该方法较传统模型具有更强的非线性拟合能力,可有效解析中药量效关系,为中药复杂系统分析提供了新的定量建模方法。
Aiming at the problem that it is difficult to analyze the multi-dimensional,small sample,nonlinear action mechanism and complex relationships between components of TCM,this paper proposes a Partial Least Squares Analysis method that integrates network features.By constructing the feature network of TCM,using the collaborative-antagonistic relationship of Louvain community recognition components,combining Moreau-Yosida regularization and Nesterov optimization to achieve feature screening,an improved Partial Least Squares Regression(PLSR)model is established to extract nonlinear components.It is verified in 8 sets of UCI datasets and Maxing Shigan Decoction cough,asthma and fever experiments.The results show that the method has stronger nonlinear fitting ability than the traditional model,which can effectively analyze the dose-effect relationship of TCM and provide a new quantitative modeling method for the analysis of complex systems of TCM.
作者
解翔杰
李欢
杨诗卓
张亚微
XIE Xiangjie;LI Huan;YANG Shizhuo;ZHANG Yawei(College of Computer Science,Jiangxi University of Chinese Medicine,Nanchang 330004,China)
出处
《现代信息科技》
2025年第13期124-128,共5页
Modern Information Technology
基金
江西省教育厅科学技术研究项目(GJJ211256)
江西中医药大学2024年大学生创新创业训练计划项目(202410412205)。
作者简介
解翔杰(2003-),男,汉族,山东烟台人,本科在读,研究方向:人工智能、数据挖掘;通信作者:李欢(1995-),女,汉族,江西萍乡人,讲师,硕士研究生,研究方向:数据挖掘、中医药信息学;杨诗卓(2004-),女,汉族,江西南昌人,本科在读,研究方向:人工智能、数据挖掘;张亚微(2005-),女,汉族,江西南昌人,本科在读,研究方向:人工智能、数据挖掘。
