环球中医药2026，Vol.19Issue(3)：428-436,9.DOI:10.3969/j.issn.1674-1749.2026.03.004
基于人机协同的名医核心处方挖掘模式探索——以荨麻疹为例
Research on core prescriptions for diseases through the integration of machine learning and expert interviews:a case study of urticaria
摘要
Abstract
Objective Constructing a human-computer collaborative research paradigm of"expert guidance-algorithm mining-expert verification"by exploring the core prescription mining of Professor Song Ping in treating urticaria.Methods The"deficiency-excess"syndrome differentiation framework for urticaria treatment was established through expert interviews with Professor Song.Outpatient medical record data of urticaria cases treated by Professor Song were extracted.Machine learning models,including Lasso regression,Random Forest,Support Vector Machine,and Gradient Boosting Machine,were employed to identify syndrome-specific single herbs for deficiency and excess syndromes.The FP-Growth algorithm was used to mine frequent herb pairs,while the Louvain and K-means algorithms were applied to identify core prescription structures.The results generated by machine learning were submitted to Professor Song for evaluation and refinement.Results The syndrome-specific herbs were identified by the Random Forest model(16 herbs for deficiency syndrome and 11 herbs for excess syndrome)showed that the highest concordance with Professor Song's clinical reasoning.Herb pairs ranked by lift,such as Wuweizi-Wumei(Schisandra Chinensis Fructus-Mume Fructus)and Mudanpi-Xuchangqing(Moutan Cortex-Cynanchi Paniculati Radix et Rhizoma),best reflected her compatibility principles.The core prescription structures uncovered by the Louvain algorithm provided a more comprehensive reflection of her academic expertise.Ultimately,the core prescription for deficiency syndrome was determined to be Shengyang Yiwei Decoction,and for excess syndrome,a self-composed prescription Urticaria Formula.Key herbs capable of treating both deficiency and excess syndromes,such as Yinchaihu(Stellariae Radix)and Xuchangqing(Cynanchi Paniculati Radix Et Rhizoma),were identified.Conclusion By taking the mining of Professor Song Ping's core prescriptions for urticaria as an example,this study constructed and practiced a human-computer collaborative research paradigm of"expert guidance-algorithm mining-expert verification",thereby providing a referential research methodology for the experience inheritance of the Xin-De School in dermatology and surgery.关键词
机器学习/专家访谈/经验传承/荨麻疹/人机协同Key words
machine learning/expert interview/experience inheritance/urticaria/human-computer collaborative分类
医药卫生引用本文复制引用
金秋百,冷学明,骆长永,段行宇,吴嘉荣,刘明玥,宁博彪,宋坪..基于人机协同的名医核心处方挖掘模式探索——以荨麻疹为例[J].环球中医药,2026,19(3):428-436,9.基金项目
北京市高层次创新创业人才支持计划"登峰"项目(项目编号G202514020) （项目编号G202514020）
世界中医药科技专项(WFCMS2025002) （WFCMS2025002）
