[Prescription and medication regularity of traditional Chinese medicine for treating child pneumonia based on data mining]

To analyze and summarize formulae of traditional Chinese medicine(TCM) in the treatment of pediatric pneumonia in CNKI based on the data mining method. CNKI was retrieved for literatures of TCM treatment of pneumonia in children in recent 20 years, which was taken the data source. After screening, a prescription database was established. Frequency analysis, association rules Apriori algorithm and complex system entropy clustering analysis methods integrated in TCMISS(V2.5) were applied in data analysis and mining. Core drugs and their properties and flavors, medication modes and new prescriptions were summarized and studied. A total of 408 prescriptions were screened out, involving 218 drugs, among which Chinese herbal medicines with the highest use frequency included Gancao(Glycyrrhizae Radix et Rhizoma), Xingren(Armeniacae Semen Amarum), Huangqin(Scutellariae Radix), Mahuang(Ephedrae Herba), Jiegeng(Platycodonis Radix), Banxia(Pinelliae Rhizoma), Sangbaipi(Mori Cortex), Chenpi(Citri Reticulatae Pericarpium), Fuling(Poria), Jinyinhua(Japonicae Lonicerae Flos). Flavors were mainly bitter, sweet and pungent; and the main medicinal property was cold. Most of these drugs entered lung meridian(LU), stomach meridian(ST) and spleen meridian(SP); and the association rules between drugs were established. Totally 6 new prescriptions were obtained through entropy hierarchical cluster analysis. Our results objectively presented, at present, TCM focuses on phlegm and heat in treating children's pneumonia, which targets at lung and also treats spleen and stomach. The main therapies aim to clear heat and resolve phlegm, ventilate lung and relieve cough and dyspnea, release exterior and dissipate cold, and relieve cough and eliminate phlegm. The results of statistical analysis are basically consistent with the consensus of experts, which can provide theoretical basis for clinical medication and drug development.