目的 应用数据挖掘技术分析含药食同源菟丝子的保健食品、中成药和方剂的组方规律，获得具有缓解疲劳和增强免疫力功能的潜在配方。方法 借助国家市场监督管理总局特殊食品信息查询平台和药智网平台收集并组建含菟丝子的保健食品、中成药和方剂的数据库。利用复杂网络分析、Apriori关联规则算法、层次分析法（analytic hierarchy process，AHP）-基于指标相关性的指标权重确定方法（criteria importance though intercrieria correlation，CRITIC）组合赋权算法和自组织映射（self-organizing maps，SOM）聚类算法从单味药、药对和药组层面研究其组方规律并设计新配方，利用逼近理想解法（technique for order preference by similarity to an ideal solution，TOPSIS）和秩和比法（rank-sum ratio，RSR）的模糊联合对配方进行排序和评价。结果 收集到45个保健食品、59个方剂、103个中成药，共142种原料，Apriori关联规则获得39对高频药对，SOM聚类将不同权重原料聚为8类。结合中医药传统理论，共获得28个潜在配方，根据模糊集理论，优化获得最优配方为菟丝子-熟地黄-杜仲-黄芪。结论 以药食同源菟丝子为例，以缓解疲劳和增强免疫力功能为导向，结合数据挖掘技术和中医药传统理论，得到最优保健食品组，为中药类保健食品的配方设计提供新思路。

Objective To analyze the formula rules of health food, Chinese patent medicines and prescription containing Tusizi (Cuscutae Semen), and obtain potential formulae with the function of relieving fatigue and enhancing immunity by data mining technology. Methods With the help of the special food information query platform of the State Administration of Market Supervision and Administration and the platform of Pharmaceutical Intelligence Network, the database of health food, traditional Chinese patent medicines and prescriptions containing Cuscutae Semen was collected and established. Using the complex network analysis, Apriori association rule algorithm, analytic hierarchy process (AHP)-criteria importance through inter criteria correlation (CRITIC) combination weighting algorithm, and self-organizing maps (SOM) clustering algorithm were carried out at the level of individual drugs, drug pairs, and drug groups, to study their formula rules and design new formulae. Using the fuzzy combination of technique for order preference by similarity to an ideal solution (TOPSIS) and rank sum ratio (RSR) methods to sort and evaluate formulae. Results A total of 142 kinds of raw materials were collected from 45 health foods, 59 prescriptions, and 103 Chinese patent medicines. A total of 39 pairs of high-frequency drug pairs were obtained from Apriori association rules and eight categories of the raw materials with different weights were clustered by SOM clustering. Based on the theory of traditional Chinese medicine, a total of 28 potential formulae were obtained. According to the Fuzzy set theory, the optimal formula was obtained as Cuscutae Semen-Shudihuang (Rehmanniae Radix Praeparata)-Duzhong (Eucommiae Cortex)-Huangqi (Astragali Radix). Conclusion This study took the medicinal and food homologous Cuscutae Semen as an example, with a focus on health functions of relieving fatigue and enhancing immunity, and combined data mining technology and traditional Chinese medicine theory to obtain the optimal formula, so as to provide new ideas for the formula design of traditional Chinese medicine health food.

G350；R285

聊城市重点研发计划项目（2021ZD07）；泰山产业领军人才工程资助（NO.tscx202211148）