基于BERTopic的中药治疗眼科疾病的处方用药规律分析  

作　　者：卢昕怡 李弘辰 吴双 罗杰[1] 

机构地区：[1]浙江中医药大学,310053

出　　处：《浙江临床医学》2024年第6期899-901,共3页Zhejiang Clinical Medical Journal

摘　　要：目的分析中医药治疗眼科疾病的处方用药规律,探讨治疗眼科疾病的更多潜在配伍组合。方法收集眼科专家门诊的中医处方并整理筛选,采用Python 3.9.10中BERTopic算法对复发性虹膜睫状体炎的所有处方进行分析,得到药物关键词聚类组合;采用SPSS 25.0对中药处方数据进行层次聚类,比较两种方法的优劣和药物配伍组合的挖掘效果。结果采用BERTopic算法得到核心药物关键词组合累计2种,包括酒地龙、土茯苓、大青叶、葎草等;应用层次聚类得到的药物组合包括葎草、土茯苓、菝葜等。应用BERTopic技术的算法具有不易受噪声数据影响、提高聚类效率、增强对于处方文本语义理解等诸多优势。结论使用基于BERTopic技术的算法在寻找疾病潜在中药配伍中表现良好,潜力较大,能为眼科疾病的中药配伍组合提供更多参考方向。Objective To analyze the prescription medication rules of tradition Chinese medicine(TCM) in the treatment of ophthalmic diseases,exploring more potential compatibility combinations of ophthalmic diseases. Methods The prescriptions of TCM in the outpatient department of ophthalmology were collected and sorted out. The BERTopic algorithm in Python 3.9.10 was used to analyze all the prescriptions of recurrent iridocyclitis, and the clustering combination of drug keywords was obtained. Also,SPSS 25 was used for hierarchical clustering of TCM prescription data to compare both advantages and disadvantages of BERTopic, exploring the mining effect of drug compatibility combinations. Results Using BERTopic has obtained 2 combinations of core drugs, including "processed pheretima asiatica, glabrous greenbrier rhizome, indigowoad leaf, humulua scandens", etc. The drug combinations obtained by hierarchical clustering including "humulua scandens, glabrous greenbrier rhizome, chinaroot greenbrier", etc. BERTopic has many advantages, including being less susceptible to noise data, improving clustering efficiency, and enhancing semantic understanding of prescriptions. Conclusion The algorithm based on BERTopic has performed well in finding potential compatibility relationship of TCM. It can lead more directions for the compatibility combination of TCM.

关 键 词：数据挖掘 用药规律 眼科疾病 BERTopic模型 

分 类 号：R276.7[医药卫生—中医五官科学]