首页|融合多维度相似算法的中药复方个性化煎煮推荐研究

融合多维度相似算法的中药复方个性化煎煮推荐研究

扫码查看
原文链接
  • 万方数据
  • 维普
目的:为实现中药汤剂智能煎煮设备煎煮参数快速推荐,提出一种融合方剂多维度相似算法,借鉴古代经典名方的个性化煎煮推荐.方法:选取30首经典名方,给出算法推荐的煎煮参数值,比较与典籍记载对应的煎煮参数值之间的差异.针对中药复方的组成成分、剂量以及功用维度特性,分别采用Jaccard相似系数、夹角余弦相似度和潜在狄利克雷分布(LDA)主题模型相似度算法,计算实验方剂与煎煮数据库中方剂间的相似度,将各维度最为相似方剂的煎煮参数加权融合,获取30首经典名方的个性化煎煮参数.结果:加水量、武火及文火时长参数值之间的相对误差分别为5.3％、3.4％、7.1％.结论:2组煎煮参数值较为接近,较好地借鉴和传承经典名方煎煮工艺.
Personalized Decoction Recommendation of Traditional Chinese Medicine Compound Formulas with Algorithm of the Multi-dimensional Similarities
Objective:To achieve rapid recommendation of decoction parameters for intelligent decoction equipment of traditional Chinese medicine(TCM)decoction,a multi-dimensional similarity algorithm integrating classical formulas was proposed to provide personalized decoction recommendations by drawing on ancient classical prescriptions.Methods:Thirty classical prescriptions were selected,and the decoction parameters recommended by the algorithm were compared with those recorded in ancient texts.The Jac-card similarity coefficient,cosine similarity,and Latent Dirichlet Allocation(LDA)topic model similarity algorithms were employed to calculate the similarity between experimental formulas and those in the decoction database,based on the composition,dosage,and functional characteristics of TCM compound formulas.The decoction parameters of the most similar formulas in each dimension were weighted and integrated to obtain personalized decoction parameters for the thirty classical prescriptions.Results:The relative errors in water volume,strong fire duration,and mild fire duration were 5.3％,3.4％,and 7.1％,respectively.Conclusion:The two sets of decoction parameters were relatively close,effectively drawing on and inheriting the decoction techniques of classical prescrip-tions.

Traditional Chinese medicine decoctionPersonalized decoctionMulti-dimensionLDA topic modelFormula simi-larityCosine of included angleJaccard similarity coefficient

李智彪、江民财、赵化勇、杜建强、熊旺平、罗计根、伍振峰

展开 >

江西中医药大学计算机学院,南昌,330004

江西中医药大学现代中药制剂教育部重点实验室,南昌,330004

江西中医药大学创新药物与高效节能降耗制药设备国家重点实验室,南昌,330004

中药汤剂 个性化煎煮 多维度 潜在狄利克雷分布主题模型 方剂相似度 夹角余弦 Jaccard相似系数

国家重点研发计划项目国家自然科学基金项目江西省教育厅科技计划项目江西省教育厅科技计划项目江西省中医药管理局科技计划项目江西中医药大学研究生创新专项

2019YFC171230182260988GJJ171452GJJ1806562022B1075JZYC21S77

2024

10.3969/j.issn.1673-7202.2024.16.004

世界中医药
世界中医药学会联合会

世界中医药

CSTPCDCHSSCD北大核心
影响因子：1.266
ISSN：1673-7202
年,卷(期)：2024.19(16)