基于数据挖掘的连建伟教授运用桑菊饮加减治疗温病组方规律研究OA

[Objective]To investigate Professor LIAN Jianwei's medication patterns in the treatment of warm disease(Wenbing)by using Sangju Yin,and to elucidate his academic thought through data mining.[Methods]A total of 534 medical records involving Sangju Yin prescribed by Professor LIAN from August 2003 to July 2013 were collected.The case data were screened and organized.Frequency analysis was conducted to evaluate the usage frequency of the original formula and high-frequency added herbs,along with their properties(taste and meridian entry).Hierarchical cluster analysis(Ward's method;Jaccard coefficient)was used to identify functional clusters of high-frequency added herbs.Apriori algorithm was applied to explore symptom-herb associations,with a focused analysis of Danggui-related rules.Phi coefficients,multivariable logistic regression[odds ratio(OR)],and random-forest feature importance were used to validate key medication indications and the diagnostic progression pathways.[Results]Frequency analysis indicated that,in the treatment of warm disease with Sangju Yin,acrid-cool herbs were frequently used,and the high-frequency added herbs followed the common medicinal properties in warm disease treatment.Cluster analysis revealed stable herb-pairing patterns.Symptom-herb association rule analysis showed that symptoms such as sore throat and cough were closely related to the use of high-frequency added herbs in Sangju Yin.Multi-model validation revealed that"tongue petechiae/ecchymosis"was a core indication for using Danggui(OR=25.97,95%CI:9.30-71.50);when wind-cold symptoms co-occurred with tongue stasis,the confidence for using Danggui reached 90.91%.[Conclusion]Professor LIAN's clinical use of Sangju Yin was characterized by adherence to a stable base-formula structure in line with WU Jutong's principle of"acrid-cool,light medicinals",while employing patterned herb clusters to address concomitant pathogenetic factors.The use of Danggui was primarily guided by tongue-stasis indicators and dynamically adjusted according to the accumulation of syndrome elements.This results provides a data-driven example for mining the tacit clinical experience of renowned senior traditional Chinese medical practitioners.