目的 利用先进数据挖掘技术探讨国医大师张磊治疗妇科疾病的用药规律和临床经验。方法 从国医大师张磊2006年至2017年11038份临床真实病例建立的医案数据库中抽取妇科月经病临床医案243条，根据排除纳入标准，得到205条完整医案。建立张磊治疗妇科月经病的医案数据库，利用Python语言可视化技术对医案数据进行科学计算和可视化输出。结果 共纳入标准医案205例，处方205首。共涉及中药176味，累计用药频次2468次，平均用药频次为14，最多使用频次为141次，频数≥20次的中药有40味，其中当归、白芍、甘草是处方中使用频次最高的3味中药。通过对处方中药进行性味归经统计：药性主要以温性、平性和寒性为主；药味以甘、苦、辛为主；药物归经以肝、肾、脾为主。基于HBE-Apriori算法对高频药物进行关联分析：置信度最高的药对依次为“川芎—当归”、“桃仁—红花”、“柴胡—当归”；置信度最高的3味药组合为“柴胡and茯苓—白芍”、“茯苓and白芍—当归”、“柴胡and茯苓—当归”；置信度最高的4味药组合为“柴胡and当归and茯苓—白芍”、“柴胡and白芍and甘草—茯苓”、“白芍and茯苓and甘草—当归”。利用Canpoy-K-means聚类算法和频繁闭项集压缩算法，对张磊教授治疗月经病的临证遣药规律进行研究挖掘。结论 可视挖掘技术可以充分而且有效协助挖掘张磊教授治疗妇科月经病的特色经验，具有较高的临床参考和学习价值。

Objective To discuss the medication rule and clinical experience of Chinese medicine master Zhang Lei in the treatment of gynecological diseases by using advanced data mining technology.Methods 243 cases of gynecological menstrual diseases were selected from the database of 11038 real clinical cases established by Chinese medical master Lei Zhang from 2006 to 2017. According to the exclusion and inclusion criteria, 205 complete cases were obtained. Establish a medical record database for Zhang Lei''s treatment of gynecological menstrual disease, and use the visualization technology of Python language to scientifically calculate and visualize the medical record data.Results A total of 205 standard medical cases and 205 traditional Chinese medicine prescriptions were included. A total of 176 Chinese medicines were involved, and the cumulative medication frequency was 2,468, with an average of 14. The most frequently used Chinese medicines were 141, and 40 were those with a frequency≥20. The three most frequently used Chinese medicines were Radix Angelicae Sinensis, Radix Paeoniae Alba and Radix Glycyrrhizae, respectively. Based on the statistical analysis of the progressive taste of traditional Chinese medicine, the medicinal nature was mainly warm, neutral and cold. The medicine flavor mainly sweet, bitter and pungent; The main channels tropism of drugs were liver, kidney and spleen. For correlation analysis of high-frequency drugs based on HBE-Apriori algorithm, the drug pairs with the highest confidence were "Rhizoma Ligustici Chuanxiong - Radix Angelicae Sinensis", "Semen Persicae - Flos Carthami" and "Radix Bupleuri - Radix Angelicae Sinensis" respectively. The three drug combinations with the highest confidence were "Radix Bupleuri and Poria - Radix Paeoniae Alba", "Poria and Radix Paeoniae Alba - Radix Angelicae Sinensis", "Radix Bupleuri and Poria - Radix Angelicae Sinensis". The four combinations with the highest confidence were "Radix Bupleuri and Radix Angelicae Sinensis and Poria - Radix Paeoniae Alba", "Radix Bupleuri - Radix Paeoniae Alba and Radix Glycyrrhizae - Poria" and "Radix Paeoniae Alba - Poria and Radix Glycyrrhizae - Radix Angelicae Sinensis". Canpoy-K-means clustering algorithm and frequent closed item set compression algorithm were used to study and explore the rule of dispensing drugs in the treatment of menstrual diseases by Professor Zhang Lei.Conclusion Visual mining technology can fully and effectively assist in mining Professor Zhang Lei''s characteristic experience in the treatment of gynecological menstrual diseases, which has high clinical reference and learning value.

R249

1.科学技术部国家重点研发计划课题(NO：2017YFC1703504):基层中医馆健康信息平台的研发与实施,负责人：张磊；2.中央级公益性科研院所基本科研业务费专项资金资助-中国中医科学院中医药信息研究所自主选题（No：ZZ120318-W）：皮肤病辅助开方系统构建；负责人：张磊；3. 河南省科技厅重点研发与推广专项（科技攻关）项目（212102311140）：国医大师张磊脾胃病验案智能语义检索系统的构建与研究应用，负责人：孙玉信。