《甘露四部》治疗赫依病方剂用药规律挖掘及知识图谱构建

    Medication Pattern Mining and Knowledge Graph Construction for Prescriptions Treating Heyi Disease in The Four Nectar Treatises

    • 摘要:
      目的 挖掘蒙医三大经典之一《甘露四部》中治疗赫依病的方剂用药规律,并构建知识图谱,为赫依病的临床用药、新药研发提供参考。
      方法 人工提取《甘露四部》中全部治疗赫依病的方剂,并按纳入、剔除及拆分规则筛选;依据《蒙医方剂全书》《中国蒙药名词规范》《中华人民共和国卫生部药品标准·蒙药分册》《蒙医学蒙汉名词术语词典》《蒙医常用经典方剂》标准化蒙药及其药味、药性、功效,采用Excel建立原始数据库;采用Excel数据透视表进行频次统计,采用SPSS Modeler开展关联规则分析,采用Jaccard相似系数与类平均法进行层次聚类分析;定义方剂、单药、功效三类实体与对应语义关系,基于Python+Neo4j构建《甘露四部》中赫依病方剂的知识图谱。
      结果 筛选《甘露四部》中镇赫依病方剂44首,涉及单味药89种,药物总出现频次387次;频次≥10次的高频镇赫依药为肉豆蔻、干姜、诃子、阿魏、荜茇、光明盐、石榴、木香、肉桂;共现频次居前的核心药对为肉豆蔻-诃子、肉豆蔻-阿魏、石榴-荜茇、干姜-荜茇、光明盐-干姜、干姜-肉豆蔻;把全部方剂划分为8个特征聚类簇(其中第1类含方剂17首,第2类含方剂13首,第3类含方剂4首,第4类含方剂4首,第5类含方剂2首,第6类含方剂2首,第7类含方剂1首,第8类含方剂1首);药物共现网络纳入35味药物,知识图谱含单药、方剂、功效三类实体,直观呈现“单药-方剂-功效”关联网络,可为核心配伍检索与潜在新方推演提供支撑。
      结论 《甘露四部》治疗赫依病的方剂高频单味药依次为肉豆蔻、干姜、诃子、阿魏、荜茇、光明盐、石榴、木香、肉桂,主要以辛味、温性、腻性的镇赫依药为核心,常配伍温胃消食、调和三根;数据挖掘结合知识图谱可量化、可视化蒙医古籍隐性用药规律,为蒙医经典数字化传承提供可行技术路径。

       

      Abstract:
      OBJECTIVE To mine the medication patterns of prescriptions for treating Heyi disease in Ganlu Sibu(The Four Nectar Treatises), one of the three classic works of Mongolian medicine, and to construct a knowledge graph, thereby providing references for clinical medication and new drug development for Heyi disease.
      METHODS  Prescriptions for Heyi disease were manually retrieved from The Four Nectar Treatises and screened according to the inclusion, exclusion and splitting rules. Mongolian medicines and their medicinal flavors, natures and efficacies were standardized with reference to Complete Book of Mongolian Medical Prescriptions, Standard Nomenclature of Mongolian Medicines, Drug Standards of the Ministry of Health of the People’s Republic of China · Mongolian Medicine Volume, Mongolian-Chinese Dictionary of Medical Terms and Classic Prescriptions Commonly Used in Mongolian Medicine, and a raw database was established in Excel. Frequency statistics were performed with Excel PivotTable, association rule analysis with SPSS Modeler, and hierarchical cluster analysis with the Jaccard similarity coefficient and the average linkage method. Three types of entities—prescriptions, single medicinal herbs and efficacies—as well as their corresponding semantic relations, were defined, and a knowledge graph of Heyi disease prescriptions in The Four Nectar Treatises was constructed based on Python and Neo4j.
      RESULTS  A total of 44 prescriptions for Heyi disease were selected from The Four Nectar Treatises, involving 89 single medicinal herbs with 387 total occurrences. The high-frequency herbs for calming Heyi(frequency ≥10) were Myristicae Semen, Zingiberis Rhizoma, Chebulae Fructus, Ferulae Resina, Piperis Longi Fructus, Halite, Pomegranate, Aucklandiae Radix and Cinnamomi Cortex. The core medicinal pairs with the highest co-occurrence frequencies were Myristicae Semen-Chebulae Fructus, Myristicae Semen-Ferulae Resina, Pomegranate-Piperis Longi Fructus, Zingiberis Rhizoma-Piperis Longi Fructus, Halite-Zingiberis Rhizoma and Zingiberis Rhizoma-Myristicae Semen. All prescriptions were divided into 8 characteristic clusters(cluster 1 contained 17 prescriptions, cluster 2 contained 13, cluster 3 contained 4, cluster 4 contained 4, cluster 5 contained 2, cluster 6 contained 2, cluster 7 contained 1 and cluster 8 contained 1). The drug co-occurrence network included 35 medicinal herbs, and the knowledge graph contained three types of entities—single herbs, prescriptions and efficacies—intuitively presenting the “single herb-prescription-efficacy” association network and providing support for the retrieval of core compatibilities and the deduction of potential new prescriptions.
      CONCLUSION In the Mongolian medical ancient book The Four Nectar Treatises, the high-frequency single medicinal herbs for treating Heyi Disease are Myristicae Semen, Zingiberis Rhizoma, Chebulae Fructus, Ferulae Resina, Piperis Longi Fructus, Halite, Pomegranate, Aucklandiae Radix and Cinnamomi Cortex in sequence. Prescriptions mainly take acrid, warm and greasy herbs for suppressing Heyi as the core, and are often combined with herbs that warm the stomach, promote digestion and harmonize the Three Roots. The combination of data mining and knowledge graph can quantify and visualize the implicit medication rules hidden in ancient Mongolian medical books, providing a feasible technical path for the digital inheritance of Mongolian medical classics.

       

    /

    返回文章
    返回