名老中医经验数据挖掘文献计量分析

目的对名老中医经验数据挖掘文献计量分析,从中挖掘规律,可指导临床,使名老中医经验知识的快速发现成为可能。方法通过系统性检索我国名老中医经验数据挖掘相关文献,运用文献计量学方法对其研究热点和内容的变化规律进行分析。结果我国名老中医经验数据挖掘相关标准文献数量为183篇,2005年以来文献的总体趋势为逐年上涨,2017年最多,占26.23%,2013年之前文献数量较少;2005-2017年文献数量平均年增长率138%。文献增长基本符合指数模型。基金资助方面,2005-2017年发表的183篇文献中共计有141篇有基金资助。2013年之前因文献数量较少,基金资助比例波动较大。2013年之后总体呈上升趋势,均在70%以上。就资助来源来看,2005-2017年国家级基金资助比例占22.95%,省部级占46.99%,合计128篇,占69.94%。2013年以来有国家级基金和省部级基金资助的文献篇数均在逐年上升。其中省部级来源的基金资助迅速增长。从主要数据挖掘工具情况来看,中医传承辅助系统占比最高,有80篇,占43.72%。其次为SPSS和Weka,分别有34篇和12篇,分别占18.58%和6.56%。另有7篇文献没有提及所用的数据挖掘工具,占3.83%。从主要统计方法的种类来看,有1篇文献应用了3种统计方法,包括关联规则、聚类分析和因子分析,占文献总数的0.55%,83篇文献应用了2种统计方法,以关联规则和聚类分析为主,占45.36%,99篇文献只用了1种统计方法,占54.1%。从具体的统计方法来看,使用关联规则文献最多,有135篇,占50.37%,聚类分析次之有99篇,占36.94%。关联规则与聚类分析联合应用最为常见的有75篇,占40.98%,其共现比例占所有统计方法共现比例的87.36%。数据来源方面,共有20篇文献涉及对历史名中医的医案或医籍进行挖掘,占10.93%,剩余163篇文献是对当代名中医经验的挖掘,占89.07%,其中126篇是对处方数据的挖掘,占68.85%,37篇是对相关文献中方剂数据进行挖掘,占20.22%。挖掘案例数量分布方面,有1篇文献没有明确的挖掘案例数量信息。其中101~200例占比最高,有61篇文献,占33.33%,其次为51~100例和201~300例占13.11%,各有24篇。统计内容方面,目前有两种研究比较常见,一是对某一类疾病的用药规律进行分析,有160篇,占87.43%;一是对某一类方剂的用药规律进行分析,有13篇,占7.1%。结论在未来的数据挖掘中应该充分的考虑中医谴药组方的规律,重点挖掘名老中医针对具体病症对经方进行加减运用的经验及其“病-证-方-药-量”的多维关联。

Bibliometric analysis on the empirical data mining of the experiences of famous and veteran TCM physicians

Objective To conduct the empirical data mining and explore the rules of the famous and veteran TCM physicians by using bibliometric analysis so as to provide the references to clinical practice and accelerate the discovery of the experiences of the famous and veteran TCM physicians.Methods The relevant articles of the famous and veteran TCM physicians were retrieved.Using the biliometric method,the changing rules of the research hotspots and contents were analyzed.Results A total of 183 eligible articles were collected.Since 2005,the article numbers were increasing by year and they were maximal in 2017 accounting for 26.23% and the numbers were less before 2013.From 2005 to 2017,the average annual growth rate was 138%.The article increase is basically coincident with the index model.Regarding the fund grant,from 2005 to 2017,141 of 183 articles were supported by the fund.Because of the few articles before 2013,the proportion of fund grant was fluctuated widely.After 2013,the general situation was increasing,over 70%.Regarding the fund source,from 2005 to 2017,the national fund grant accounted for 22.95%,the fund from the province and ministry projects for 46.99 %,with 128 articles in total,accounting for 69.94%.Since 2013,the articles supported by the fund of the national,province and ministry levels were increasing by years.Of them,the fund from the province projects was dominated.Regarding the data mining tool,80 articles were analyzed with TCM inheritance adjuvant system,accounting for 43.72%,34 articles with SPSS and 12 articles with Weka,accounting for 18.58% and 6.56% separately.The other 7 articles did not mention the data mining tool,accounting for 3.83%.Regarding the statistical methods,1 articles used three methods,named association rule,clustering analysis and factor analysis,accounting for 0.55%.83 articles used 2 statistical methods,named association rule and clustering analysis,accounting for 45.36% and 99 articles used one statistical method,accounting for 54.1 %.In general,135 articles used association rule,accounting for 50.37%,99 articles used clustering analysis,for 36.94 %.The combination of association rule and clustering analysis were used in 75 articles,accounting for 40.98% and the co-occurrence proportion was 87.36%.Regarding the data source,20 articles were related to the medical cases or works of the ancient eminent TCM masters,accounting for 10.93% and the rest 163 articles referred to the experiences of the modern famous TCM physicians,for 89.07%.126 articles of them were related to the formula data,accounting for 68.85 %,37 articles to the formula data of the relevant articles,for 20.22%.Regarding the distribution of case numbers,1 article had no information of medical case.61 articles included 101 to 200 cases,accounting for 33.33% and 24 articles included 51 to 100 cases and 201 to 300 cases,accounting for 13.11%.Regarding the statistical content,two topics were common,one was the medication rule for a kind of diseases,including 160 articles,accounting for 87.43% and another one was the medication rule for a kind of herbal formula,including 13 articles,accounting for 7.1%.Conclusion In the future data mining,the rules of herbal composition should be fully considered,the experiences of the famous and veteran TCM physicians in the modification of the classical formulas for the individual disorders as well as the multiple correlation of disease-syndrome-formula-drug-dose should be explored deeply.