植物药的宏观药性是其物质成分微观药性整体协同呈现的结果，具有明显的系统性和非线性特征，任何单一或几种成分均不能完整表征其宏观药性，宏观药性也不是多种成分活性的简单相加。我们此前开展课题研究发现，中药全成分水提红外光谱数据SVM识别方法在对60种植物药寒热药性辨识过程中显示出明显优势，提取出了辨识中药寒热性质和寒热程度的特征参数，技术手段客观可重复。由于前期研究所选中药涵盖多科属植物，模型对60种中药的组内预测正确率虽已达100%，但对组外同样本量其他植物药的随机预测正确率仍不够高（58.3%）。基于此，我们将植物遗传亲缘关系与药性研究结合，拟以唇形科药用植物为示例，选取本科属寒热中药各20种，分析与药性关联的化学成分特征信息，力求在科属单元内，实现基于中药全成分特征标记的寒热药性识别和预测，为分析中药药性成分复杂体系问题和基于模式识别指导下的药性成分提取与验证提供科学依据。

The macroscopic properties of plant medicine is the result of together present of microscopic properties of the material composition, which has the significant systemic and nonlinear characteristics. Any single or servral component could not whole performance the macroscopic properties, and the macroscopic properties not only the sum of servral component. Our previous study shown that infrared spectral data SVM pattern has the absolute advantage in identification cold-heat nature of 60 kinds of medicinal plants, and the characteristic parameters of nature and degree of cold-heat were found. The prediction accuracy has reached 100% within the group to predict the 60 kinds of medicinal plants,but the random prediction accuracy still not tall enough (58.3%) in other medicinal plants. Based on this, we combination medicinal research with plant genetic relationships, lamiaceae medicinal plant as example. 20 kinds of medicinal plants were selected, and the information of chemical composition characteristics were analyzed. To implement identification and prediction on cold-heat nature basis on the features marked of total components, and provid scientific basis for extracting medicinal components and pharmacodynamics authentication under the guidance of the recognition pattern and analysis complex system problem of Chinese herbal medicine composition.