蒙医药清瘟杀黏类方剂组方的知识发现研究

Research on Knowledge Discovery of Mongolian Medicine Clearing the Plague and Killing Sticky Disease Prescriptions

目的 本研究对181首“清瘟杀黏”功效的方剂数据集进行数据处理与分析,挖掘蒙医治疗瘟疫组方用药规律。以内蒙古自治区卫生健康委员会发布的《新型冠状病毒肺炎蒙医预防和诊疗方案(试行第三版)》提供的治疗新冠肺炎总方药为基准,从清瘟杀黏数据集中分别为蒙医治疗新冠肺炎的未成熟热期、炽热期普通型、炽热期重症型3个阶段推荐清瘟杀黏方剂。方法 对方剂数据集药物词频与高频药物的药效、药味、药性进行数据统计;使用K-means机器学习算法对蒙医治疗瘟疫高频药物进行聚类分析;通过FPGrowth关联算法,抽取方剂中相关数据实体及各实体间关系;构建协同过滤算法程序,将药物交集最大、欧氏距离最小的方剂作为推荐结果。结果 数据集内单味药出现频率最高的是麝香、草乌、诃子、红花等。对高频药物分析统计发现,钝、轻、重效能出现次数较多,苦味和凉性药物出现频次较高,通过三维聚类分析后得到5类结果。对数据集内所有药物数据进行关联分析,得到9个频繁项集药物组合。协同过滤算法为3个阶段推荐出“敖必德斯音哇其尔-36”等6首清瘟杀黏方剂。结论 本研究将现代信息技术与蒙医传统方剂进行深度融合,开展知识发现研究。蒙医方剂治疗瘟疫组方用药规律挖掘的相关研究结果可以为蒙医药新药数字化和智能化研发提供参考,清瘟杀黏方剂推荐用药方法的构建可以为蒙医临床治疗COVID-19精准施治以及相关科研工作提供思路。同时,本研究希冀助力蒙医传统方剂的弘扬与现代化发展。

Objective To conduct data processing and analysis on the data set of 181 prescriptions with the efficacy of "clearing plague and killing stickiness", and dig out the rules of Mongolian medicine treatment of plague prescriptions.Based on the general prescriptions for the treatment of new coronary pneumonia provided by the New Coronavirus Pneumonia Prevention and Diagnosis and Treatment Plan(Third Edition) issued by the Inner Mongolia AutonomousRegion Health Commission, the Mongolian medicine treatment of new coronary pneumonia from the clearing the plagueand killing sticky disease data set is used as a benchmark. The three stages of the immature heat stage, the common typein the hot period, and the severe type in the hot period are recommended to clear the wind and anti-sticking prescription.Methods We completed the statistics of word frequency and of drug efficacy, drug taste and drug potency of highfrequency drugs, realized clustering analysis of high-frequency drugs for plague treatment in Mongolian medicine byusing K means machine learning algorithm. By FP-Growth association algorithm, we extracted the relevant data entitiesin the prescription and the relationship between the entities. We constructed a collaborative filtering algorithm program,and used the prescription with the largest drug intersection and the smallest Euclidean distance as the recommendationresult.Results In the data set, the single herbs with the highest frequency were Moschus, Radix Aconiti Kusnezoffii,Fructus Chebulae, Flos Carthami, etc. According to the statistics of high-frequency drug analysis, there were morefrequent occurrences of blunt, light, and heavy efficacy, and higher frequency of bitter and cool drugs. Five types ofresults were obtained after three-dimensional cluster analysis. We performed correlation analysis on all drug data in thedata set, and obtained 9 frequent item set drug combinations. The collaborative filtering algorithm recommended 6 antisticking prescriptions such as "Aobidesiyinwaqier-36" for 3 stages.Conclusion This research deeply integrates moderninformation technology and traditional Mongolian medicine prescriptions to carry out knowledge discovery research. Therelated research results of Mongolian medicine prescriptions for the treatment of plague prescriptions can provide areference for the digital and intelligent research and development of new Mongolian medicines. The construction of therecommended drug method of clearing the plague and killing sticky disease prescription can provide precise treatmentfor the clinical treatment of COVID-19 in Mongolian medicine. It provides ideas for related scientific research work. Atthe same time, this research hopes to help the promotion and modernization of traditional Mongolian medicineprescriptions.