自组织映射模型在医学领域数据挖掘中的应用  

Applications of self-organizing maps to the data mining in medical domain

作　　者：孙丽君[1] 尚磊[1] 苏海霞[2] 王锐[1] 李婵娟[1] 杨显君[1] 张玉海[1] 

机构地区：[1]第四军医大学卫生统计学教研室,陕西省西安市710032 [2]第四军医大学流行病学教研室

出　　处：《中国医院统计》2011年第3期198-201,共4页Chinese Journal of Hospital Statistics

基　　金：国家自然科学基金（30901241）陕西省科技攻关项目（2009K18-01）

摘　　要：目的 介绍自组织映射模型的基本原理、常见算法及其研究进展,阐述自组织映射模型在医学领域数据挖掘中的应用,以期为相关研究提供参考方法 本文从医学领域的数据挖掘角度出发,采用文献回顾的方法,阐述自组织映射模型在处理非线性、高度冗余数据时运算速度和精度上的优势,并分析自组织映射模型各种算法的特点及不足结果 相对于传统的数据挖掘方法在大规模的医学数据挖掘时无法消除数据冗余及捕捉非线性特征,从而导致挖掘过程速度慢、精度低等问题,自组织映射模型可以自动地、智能地、可视化地将待处理的数据转化为有用的知识,较好地弥补传统数据挖掘方法的不足.自组织映射模型算法的不断改进,保证了它可以适用于具有不同特征的数据结论 自组织映射模型在医学领域的数据挖掘中得到了较好的应用,其算法还有进一步改进的空间,在医学领域的应用前景将会得到进一步扩展.Objective To introduce the applications of self-organizing maps （SOM） to data mining in medical domain and illustrate the research development of the model and its algorithms, offer theoretical reference for the researchers in relevant medical domain. Methods From the view of data mining in medical domain, we adopted literature review method to illustrate the superiorities of SOM in operation speed and precision aspects for nonlinear and high level redundant data. We also analyzed and compared the characteristics and deficiencies of the SOM algorithms. Results The traditional data mining methods encounter the problems of slow operation speed and low precision for nonlinear and high level redundant data. On the contrary, SOM can transform data to useful knowledge automatically and intelligently in a visualization mode. SOM can make up the deficiencies of traditional data mining methods to some extent. Continues improvement of the SOM algorithms ensure that it can be applied to the data with different characteristics. Conclusion SOM has been widely applied to data mining in medical domain. The algorithms of SOM have potential to improve and the application range of this method in medical domain will be extended.

关 键 词：自组织映射 数据挖掘 知识发现 医学领域 

分 类 号：R318.5[医药卫生—生物医学工程]