面向中医方药分析的数据挖掘方法及应用研究

发布时间：2018-05-24 07:22

本文选题：中医方药 + 数据挖掘　；参考：《南京中医药大学》2017年硕士论文

【摘要】：中医作为我国四大国粹之一,在数千年的发展过程中,形成了独特的辨证论治理论与遣药组方规律,为人类健康做出了重要贡献,是人类的宝贵财富。随着中医药现代化进程的推进,中医方药研究不断深入,中医药学与多领域多学科相互渗透,形成了多学科交叉研究的发展趋势。数据挖掘技术能够在大量的、不规范的、模糊不完全的数据集中发现具有理论或实践价值的潜在知识,是利用数据发现知识的关键技术,已被广泛应用于各学科领域。在中医药领域中应用也开始兴起,与传统的文献研究不同,数据挖掘技术完全是根据客观的中医药数据来发现模型和探索规律,不掺杂主观因素。就中医方药研究领域的应用已展现出良好的前景,为中医方药研究开辟了新的途径。论文在研究了数据挖掘在中医方药领域的研究现状的基础上,收集整理了500余条辛温解表剂原始数据,参考《中华人民共和国药典》、《中医大辞典》、《中医临证备要》和《中医症状鉴别诊断学》等文献资料对方剂药名处理、病症数据处理、剂量单位处理等数据清洗和规范整理后,获得了 282条有效数据,建立了辛温解表剂中医方药数据库。在经过聚类分析、分类分析、关联规则挖掘、预测分析等预实验研究后,确定较为可行、算法适应、结论合理的数据分析与挖掘方法对辛温解表剂数据进行分析研究。研究了数据挖掘的主要方法,重点是聚类分析k-means算法、关联规则Apriori算法和预测模型BP算法的核心思想、基本原理和算法步骤。利用k-means算法建立中医方药聚类分析模型,对辛温解表剂数据进行离群点分析,发现当聚成2类时,成员数目分别为212和5,说明辛温解表剂数据几乎都属于一个类别,也就说明辛温解表剂的药物用量遵循着一定的用量规律,正常情况下不会出现用量过多或过少的情况。利用Apriori算法对辛温解表剂数据进行关联规则挖掘,发现了辛温解表剂中的常见药物组对和药症对应关系,挖掘结果符合中医方药的理论和实践,能够为临床用药提供辅助指导。利用BP神经网络建立辛温解表剂药症关系预测模型,对辛温解表剂样本数据进行网络训练和预测,探索辛温解表剂药物组方与主治病症之间的对应关系,研究发现通过方剂药物组合确定主治病症的准确率较高,而通过主治病症选取核心单味药的准确率较低。
[Abstract]:Chinese medicine as one of the four quintessence of our country, in the development of thousands of years, has formed a unique theory of syndrome differentiation and prescription, has made an important contribution to human health, is the precious wealth of mankind. With the advancement of the modernization of traditional Chinese medicine, the research of traditional Chinese medicine has been deepened, and the interpenetration of traditional Chinese medicine and multi-discipline has formed the development trend of multidisciplinary research. Data mining technology can discover potential knowledge with theoretical or practical value in a large number of non-standard fuzzy incomplete data sets. It is a key technology to use data discovery knowledge and has been widely used in various disciplines. Different from the traditional literature research, the data mining technology is based on the objective TCM data to discover the model and explore the law completely, and does not mix the subjective factors. The application of traditional Chinese medicine has shown a good prospect and opened up a new way for the research of traditional Chinese medicine. On the basis of studying the current situation of data mining in the field of traditional Chinese medicine, the paper collects and arranges the original data of more than 500 kinds of thermopharmaceuticals. Referring to the documents such as Pharmacopoeia of the people's Republic of China, Dictionary of the University of traditional Chinese Medicine, Preparedness of Clinical Syndromes of traditional Chinese Medicine and differential Diagnostics of traditional Chinese Medicine, etc., after cleaning and standardizing the treatment of prescription drug names, disease data processing, dose unit processing, etc., 282 effective data were obtained, and the database of TCM prescriptions of Xin-Wen antitabular agents was established. After cluster analysis, classification analysis, association rule mining, prediction analysis and other pre-experimental studies, it is determined that the algorithm is more feasible and adaptive. Conclusion the reasonable data analysis and mining methods are used to analyze and study the pungent table data. This paper studies the main methods of data mining, focusing on the core idea, basic principle and algorithm steps of clustering analysis k-means algorithm, association rule Apriori algorithm and prediction model BP algorithm. The cluster analysis model of traditional Chinese medicine prescription drugs was established by k-means algorithm, and the outlier analysis was carried out on the data of xingwen alexitizer. It was found that the number of members was 212 and 5 respectively when the data were clustered into two classes, which indicated that the data of Xin-wen alexitizer almost belong to one category. It also shows that the amount of zymopharmaceuticals follows a certain amount of law, normally there will not be too much or too little situation. By using Apriori algorithm to mine the association rules of the data of the photic temperature table agent, the corresponding relationship between the common drug pairs and the drug symptom was found. The results of the mining accord with the theory and practice of traditional Chinese medicine, and can provide auxiliary guidance for the clinical use of drugs. BP neural network was used to establish the prediction model of the relationship between photic temperature and antipyretic agents, to train and predict the sample data, and to explore the corresponding relationship between the prescription and the main disease. It was found that the accuracy rate of determining the main disease by the combination of prescriptions was higher, but the accuracy of selecting the core single medicine by the main treatment was lower.
【学位授予单位】：南京中医药大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP311.13;R28

【相似文献】