目的 实现海量中药方剂的功效自动化分类，辅助建设更加健全完备的中药方剂信息化编码体系。方法 以中药方剂为样本，以组成方剂的中药材和中药饮片为特征，利用方剂主成分相似性对样本间关系进行表征，设计一种图卷积神经网络模型，并基于国家标准《中药方剂编码规则及编码》（GB/T31773—2015）的方剂分类标准，以其中1 089首经典古方剂的功效分类作为数据集进行模型训练。结果 在少量样本条件下，可以实现中药方剂所在的Top-6功效分类87.96%的预测准确率，与其他方法对比性能更占优势。以国标中的方剂功效信息进行验证，可以很好地预测中药方剂的潜在功效。结论 基于图卷积神经网络的方法可以用于中药方剂的自动化分类，帮助转变中药标准化工作中大量专业人力投入的工作模式，并辅助中药研究者发现中药方剂的潜在临床功效。

Objective To achieve automated classification of the therapeutic effects of a massive number of traditional Chinese medicine prescriptions and assist in the construction of a more robust and comprehensive information-based coding system for traditional Chinese medicine prescriptions. Methods Traditional Chinese medicine prescriptions were used as samples, with the traditional Chinese medicinal materials and traditional Chinese medicine decoction pieces comprising the features. The relationships between the samples were represented using the similarity of the main components of the prescriptions. A graph convolutional neural network model was designed, and based on the traditional Chinese medicine prescription classification standard in the national standard Coding Rules and Coding of Traditional Chinese Medicine Prescriptions (GB/T31773-2015), the therapeutic effect classification of 1 089 classic ancient traditional Chinese medicine prescriptions in the national standard was used as the dataset for model training. Results Under a limited number of samples, a prediction accuracy of 87.96% was achieved for the Top-6 therapeutic effect classifications of traditional Chinese medicine prescriptions. Comparative experiments with other methods demonstrated the significant performance advantages of the proposed method. Furthermore, a comparison with the therapeutic effect information in the national standard indicated that the proposed method can be used to predict the potential therapeutic effects of traditional Chinese medicine prescriptions. Conclusion The method based on graph convolutional neural networks can be used to automate the classification of traditional Chinese medicine prescriptions, helping to transform the work mode of extensive human resources in traditional Chinese medicine standardization and assisting traditional Chinese medicine researchers in discovering the potential clinical efficacy of traditional Chinese medicine prescriptions.

TP18;R289

吉林省自然科学基金面上项目（20240101349JC）;吉林省教育厅科学研究项目（JJKH20240947KJ）;吉林省教育科学“十四五”规划2023年度课题（GH23163）