摘要
目的:基于自适应模糊推理系统,探讨慢性呼吸衰竭常见证候及其特征。方法:收集4所三甲医院慢性呼吸衰竭患者资料300例,使用Epidata软件建立数据库。选取人工神经网络、模糊系统,运用MATLAB6.5软件进行编程。采用动态kehonen网络,通过增加动态神经元形成动态自适应神经网络,形成最优模糊规则。基于临床数据挖掘结果获取慢性呼吸衰竭证候常见证候特征,并检验其合理性。结果:通过规则转换,主次症的筛选,明确7个证候及其对应的主、次症,分别为:痰热壅肺证、痰瘀阻肺证、阳虚水犯证、痰湿壅肺证、气阴两虚证、痰瘀阻肺兼阴虚证、痰蒙神窍证。证候诊断标准检验结果诊断符合率为74.6%。结论:该方法所获取的模糊分类规则以较高的精度反映了学习样本集中存在的规律性,说明了模型的可靠性。说明了自适应模糊推理系统可用于中医证候特征的研究。
Objective: Investigate the common TCM syndromes and its characteristic of chronic respiratory failure(CRF) based on Neuro-fuzzy theory.Methods: We collected clinical data from the patients with CRF in four grade A class 3 hospitals,established a database by using the Epidata software.Then we worked out the plan for modeling: building up fuzzy neural net models on the basis of dynamic Kohonen network and testing their reliability with the Fisher-iris data.Based on clinical data mining results we obtained the common TCM syndromes and its characteristic of CRF,moreover,to test its rationality.Results: Through rules conversion,selection of primary and secondary symptoms,we defined 7 syndromes and its primary and secondary symptoms were: syndrome of phlegm-heat obstructing lung;syndrome of phlegm and blood stasis obstructing lung;syndrome of water overflowing due to yang deficiency;syndrome of phlegm-dampness obstructing lung;syndrome of deficiency of both qi and yin;syndrome of phlegm and blood stasis obstructing lung and yin deficiency;syndrome of orifices confused by phlegm.Syndrome diagnostic criteria testing results showed that diagnostic accuracy rate was 74.6%.Conclusion: Testing model through Fisher-iris data,showed that this method to obtain the fuzzy classification rules could reflect the regularity of the samples with high accuracy,showed the reliability of the model.This shows that the Neuro-fuzzy theory can be used to study on the characteristics of TCM syndrome.
出处
《中华中医药杂志》
CAS
CSCD
北大核心
2011年第12期2770-2773,共4页
China Journal of Traditional Chinese Medicine and Pharmacy
基金
河南省高校新世纪优秀人才支持计划(No.2006HANCET-05)~~