Objective To analyze the compatibility rule of Chinese herbal medicine used by professor ZHOU Ping′an in treating connective tissue disease interstitial lung diseases（CTD-ILD） through data mining. Methods The data of CTD-ILD patients who attended ZHOU′s outpatient clinic at Dongfang Hospital of Beijing University of Chinese Medicine from January 2010 to December 2016

and had at least one follow-up visit

with definitively positive effect and complete medical records

were collected

including general information

diagnosis

history of present illness

past history

tongue coating

pulse manifestation and prescriptions. Establish a database

and perform cluster analysis and implement data mining process of association rules

so as to explore the compatibility rules of Chinese herbal medicine. Results A total of 375 Chinese herbal medicine prescriptions were used

involving 212 different herbs and 18 herbal types. The cumulative frequency of herbal use was 7251 times

and 26 herbs had frequency ≥ 90; the top five most frequently used herbs were Huangqi（Radix Astragali

372 times）

Gancao（Radix et Rhizoma Glycyrrhizae

342 times）

Danggui（Radix Angelicae Sinensis

339 times）

Baizhu（Rhizoma Atractylodis Macrocephalae

268 times） and Chuanshanlong（Rhizoma Dioscoreae Nipponicae

268 times）; the top five most frequently seen herbal types were deficiency-supplementing drugs（45 times） and heat-clearing drugs（38 times）

phlegm-dissolving

cough-relieving and panting-calming drugs（24 times）

exterior-releasing drugs（17 times）

and wind-dampness-dispelling drugs（13 times）; the top five most commonly seen medicinal nature was warm（50 times）

cold（46 times）

neutral（44 times）

slightly cold（32 times）

and slightly warm（22 times）; the top five medicinal flavors were sweet（98 times）

bitter（91 times）

acrid（82 times）

slightly bitter（21 times）

and astringent（18 times）; the top five channel entries were liver channel（106 times）

lung channel（96 times）

spleen channel（69 times）

kidney channel（68 times）

and stomach channel（64 times）. The cluster analysis results of high-frequency drugs showed that the high-frequency drugs could be divided into 4 categories when the intercept was 20. The association rule analysis of high-frequency drugs indicated that the top 3 of the binomial correlations were Gancao-Huangqi

Danggui-Huangqi

and Danggui-Gancao. The top 2 of the trinomial correlations were Huangqi-Baizhu-Gancao and Huangqi-Jixueteng（Caulic Spatholobi）-Danggui. Conclusion Professor ZHOU Ping′an mainly uses sweet-warm and bitter-cold drugs for the treatment of CTD-ILD

and the treatment principle is rooted to supplementing the center and boosting qi

banking up earth to generate metal

with qi-boosting and blood-invigorating to unblock the collaterals

and heat-clearing and phlegm-dissolving to diffuse bi（痹）.