摘要
目的 中药升降浮沉药性判别,建立中药升降浮沉药性与中药成分的相关关系。方法 依据已有结论“中药寒热药性的物质基础是中药成分”提出假说:中药升降浮沉药性的物质基础也是中药成分。实验筛选54味升降浮沉药性明确的中药作为研究对象,通过紫外指纹图谱表征中药成分,应用主成分分析(PCA)提取药性相关的成分信息,引入支持向量机(SVM)算法建立具有中医药特色的中药升降浮沉药性判别模型。结果 基于交叉验证和网格寻优算法,支持向量机算法参数c和g都为0.5时模型最优,训练集分类准确率为80.77%,测试集分类准确率为80%,模型稳定性评价的ROC曲线下面积为0.832。结论 经实验验证,此实验构建的判别模型稳定性较好,判别模型可行有效。
Objective To establish the correlation between the lifting, floating and sinking properties of traditional Chinese medicine(TCM) and the components of TCM. Methods According to the existing conclusion, "the material basis of the cold and hot properties of traditional Chinese medicine is the composition of TCM", then we put forward a hypothesis: the material basis of the uplift, float and sink of traditional Chinese medicine is also the composition of traditional Chinese medicine. In this paper, 54 traditional Chinese medicines with definite properties of uplift, float and sink were selected as the research object. The components of traditional Chinese medicine were characterized by UV fingerprint, the component information related to drug properties was extracted by principal component analysis(PCA), and the support vector machines(SVM) algorithm was introduced to establish the drug property discrimination model of traditional Chinese medicine with the characteristics of traditional Chinese medicine. Results Based on cross validation and grid optimization algorithm, the model is optimal when the parameters c and g of support vector machine algorithm are 0.5. The classification accuracy of training set is 80.77%, and that of test set is 80%. The area under ROC curve of model stability evaluation is 0.832.Conclusion The experimental results show that the discrimination model constructed in this paper has good stability and is feasible and effective.
作者
油雨忻
李若轩
段梦雨
魏国辉
YOU Yu-xin;LI Ruo-xuan;DUAN Meng-yu;WEI Guo-hui(School of Intelligent and Information Engineering,Shandong University of Traditional Chinese Medicine,Ji′nan250355,China)
出处
《时珍国医国药》
CAS
CSCD
北大核心
2022年第11期2801-2804,共4页
Lishizhen Medicine and Materia Medica Research
基金
国家自然科学基金(81473369)
山东省研究生教育优质课程和专业学位研究生教学案例库立项项目(SDYAL20050)
山东中医药大学省级创新创业训练计划项目(S202110441034)。
关键词
升降浮沉
药性
紫外指纹图谱
主成分分析
支持向量机
Uplift
float and sink
Chinese medicine medicinal properties
UV fingerprint spectra
Principal component analysis
Support vector machines
