主 办:北 京 中 医 药 大 学
ISSN 1006-2157 CN 11-3574/R

北京中医药大学学报 ›› 2019, Vol. 42 ›› Issue (2): 113-119.doi: 10.3969/j.issn.1006-2157.2019.02.005

• 科技之窗 • 上一篇    下一篇

不同证候慢性心衰生物学指标神经网络建模对比研究

王娟1,赵慧辉1,陈建新1,罗良涛2,付帮泽1,朱斌1,王伟1#   

  1. 1 北京中医药大学 北京 100029;
    2 首都医科大学
  • 出版日期:2019-02-28 发布日期:2019-02-28
  • 通讯作者: 王伟,男,博士,教授,博士生导师,主要研究方向:中医药干预心血管疾病的临床和基础研究,E-mail:wangwei@bucm.edu.cn
  • 作者简介:王娟,女,博士,副研究员
  • 基金资助:
    国家重点研发计划项目(No.2017YFC1700100)

Comparative study on the biological indicators of chronic heart failure with different patterns based on neural network modeling

Wang Juan1, Zhao Huihui1, Chen Jianxin1, Luo Liangtao2, Fu Bangze1, Zhu Bin1, Wang Wei1#   

  1. 1 Beijing University of Chinese Medicine, Beijing 100029, China;
    2 Capital Medical University, Beijing 100069, China
  • Online:2019-02-28 Published:2019-02-28
  • Contact: Wang Wei, Ph.D., Doctoral supervisor. Principal's office, Beijing University of Chinese Medicine. E-mail: wangwei@bucm.edu.cn

摘要: 目的 应用CytoScape软件和神经网络挖掘方法分别构建慢性心衰(CHF)阴虚证和阳虚证多系统理化指标信息的诊断模型,探索慢性心衰阴虚证和阳虚证病人理化指标信息的组合模式及其生物学意义。方法 收集100例CHF病人四诊信息和生物样本进行多系统理化指标的检测。在应用分析差异指标基础上,综合应用CytoScape软件及神经网络数据挖掘方法进行相关性分析和数据建模,从而形成慢性心衰不同证候患者多系统理化指标的诊断模型。结果 应用上述方法筛选出阴虚证和阳虚证有统计学意义的多系统理化指标,依次进入神经网络数据挖掘并建模,每种证候都取得了较高的准确性和预测度。结论 神经网络数据挖掘方法不仅可以用于临床理化指标信息数据进行CHF病人证候的建模分析,而且还能深入挖掘和揭示与CHF不同证候相关的多系统理化指标信息,为深入揭示不同中医证候心衰的生物学基础提供参考。

关键词: 慢性心衰, 生物学指标, 神经网络, 数据挖掘

Abstract: Objective To explore the combined model of physicochemical index information of chronic heart failure (CHF) patients with different patterns and its biological significance, CytoScape software and neural network mining methods were used to construct the diagnostic model of multiple systematic physicochemical index information of yin-deficiency-pattern and yang-deficiency-pattern CHF. Methods 100 CHF patients’ four diagnostic information and biological samples were collected for the detection of multiple systematic physicochemical indexes. On the basis of the analysis of the differential indicators, the correlation analysis and data modeling were carried out with CytoScape software and data mining method of neural network to form a diagnosis model of multiple systematic physical and chemical indicators for patients with different syndromes of CHF. Results The physical and chemical indexes with statistical significance of yin deficiency and yang deficiency were screened out with the above methods; the neural network data mining and modeling were conducted accordingly. Conclusion The neural network data mining method can not only be used for the modeling and analysis of clinical physicochemical index information of CHF patients, but also can deeply explore and reveal the multi-system physicochemical index information related to different patterns of CHF, providing references for the in-depth exploration of the biological basis of different TCM patterns of CHF.

Key words: chronic heart failure, physical and chemical index, neural network analysis, data mining

中图分类号: 

  • R256.29