目的:筛选脑得生抗血栓有效部位。方法:采用尾静脉注射胶原蛋白-肾上腺素诱发小鼠体内血栓,基于正交设计,建立复方药效的支持向量机预测模型;并经实验验证,确定脑得生抗血栓的有效部位。结果:本研究所建立的复方预测模型可对复方不同组合进行药效预测。结论:通过模型预测和实验验证,多个配伍组的药效优于原方,尤以CD效果最佳。 Objective: To screen effective antithrombotic brain parts. Methods: Thrombus was induced in mice by injection of collagen-epinephrine through the tail vein. Based on the orthogonal design, a support vector machine (SVM) prediction model was established for the compound efficacy. The validated antithrombotic sites were determined by experiments. Results: The compound prediction model established in this study can predict the pharmacodynamic effects of different combinations of compounds. Conclusion: Through the model prediction and experimental verification, the efficacy of multiple compatibility groups is better than the original one, especially the CD effect is the best.