论文部分内容阅读
目的:探讨数学建模策略在中药药性物质基础研究中的应用,并尝试建立一套客观可重复的建模思路与方法。方法:从数据库建立、数据预处理、建模策略选择和评价标准制定4个关键步骤,讨论中药药性特征标记识别模型的建立。结果:从64种建模策略中选出17种判别良好的模型,继而考察其可视化表达能力选出4种适宜模型,最后经统计模拟,选出PLS作为目前最适宜模型。结论:中药药性与其发挥作用的物质基础之间难以找到现成的公式或规律来表达二者关联,可用穷举策略,经数据库建立、数据预处理、建模策略选择和评价4步筛选适宜的中药药性特征标记识别模型。
Objective: To explore the application of mathematical modeling strategies in the basic research of Chinese medicinal materials, and try to establish a set of objective and repeatable modeling ideas and methods. Methods: From the database establishment, data preprocessing, modeling strategy selection and evaluation criteria to establish four key steps to discuss the establishment of Chinese medicinal characteristics signature recognition model. Results: Seventeen well-discriminated models were selected from the 64 modeling strategies. Then 4 suitable models were selected based on their visualization ability. Finally, PLS was selected as the most suitable model by statistical simulation. Conclusion: It is difficult to find out the existing formulas or laws to express the relationship between the medicinal properties and the material basis of Chinese medicine. The method can be used exhaustively, and the suitable traditional Chinese medicine (TCM) can be screened through database construction, data preprocessing, modeling strategy selection and evaluation Pharmacological signature recognition model.