针刺手法数据挖掘的关联规则与分类

Association Rules and Classification of Acupuncture Manipulation Data Mining

目的探讨数据挖掘技术中的关联规则与分类规则在针刺手法测定中的应用。方法分别使用提插平补平泻、提插补法、提插泻法、捻转平补平泻、捻转补法、捻转泻法6种针刺手法,在ATP-II型针刺手法测定仪上进行总计200多次针刺手法操作,取得大量针刺手法参数。对测量参数整理后建立数据库;采用WEKA数据挖掘软件对其进行数据挖掘分析,主要采用关联规则中的Apriori关联规则和分类规则中的C4.5分类方法。结果 Apriori关联规则挖掘出的结果:提插手法与提插最大值、提插最小值存在一定关联并且捻转离散度最小;捻转手法的提插离散度最小并且捻转幅度较大。C4.5分类规则较好的区分了提插手法与捻转手法。对个别手法特点做出了正确的评估,提插平补平泻手法捻转离散度小,提插频率高;捻转手法的捻转离散度较高,提插频率低。结论应用数据挖掘技术在针刺手法参数的挖掘上具有可行性,也具有进一步研究的价值与应用前景。

Objective To explore the application of the association and classification rules of data mining techniques to the determination of acupuncture manipulations. Methods Six kinds of acupuncture manipulations, namely lifting-thrusting uniform reinforcement and reduction, lifting-thrusting reinforcement, lifting-thrusting reduction, twirling uniform reinforcement and reduction, twirling reinforcement and twirling reduction methods were used separately. A total of over 200 acupuncture manipulations were performed using an ATP-II acupuncture manipulation determination instrument to obtain a large number of acupuncture manipulation parameters. A database was established after the measurement parameters were sorted out. Their data mining and analysis were performed using WEKA data mining software. The Apriori association rule and C4.5 classification method were mainly applied. Results The outcome mined according to the Apriori association rule showed that lifting-thrusting manipulations formed a certain association with the maximum lifting-thrusting value and the minimum lifting-thrusting value and had the minimum twirling dispersion degree. Twirling manipulations had the minimum lifting-thrusting dispersion degree and a larger twirling range. The C4.5 classification rule distinguished better between lifting-thrusting and twirling manipulations and made a correct assessment of the characteristics of individual manipulations. Lifting-thrusting uniform reinforcing-reducing manipulation had a small twirling dispersion degree and a high lifting-thrusting frequency. Twirling manipulations had a higher twirling dispersion degree and a low lifting-thrusting frequency. Conclusion Using data mining techniques to obtain acupuncture manipulation parameters is feasible and also has the value of further research and the prospect of application.