Abstract: In the field of tumor marker, detecting some markers together can improve the successful detection score. The cloud model has great advantages in interval definition of numerical attribute. By fuzzing the boundary of attribute and membership degree, the cloud model can express the real world well. This paper presents a combined markers detection discovery method to mine association and find out most valuable tumor markers in examination dataset. The experiments show 80% of the top 10 combined markers are fit for iatrical knowledge.

Key words: data mining, association rule, cloud model, tumor marker, combined detection

摘要： 医学发现，将具有相关性的肿瘤指标进行联合检测能提高癌症的阳性检出率。基于云模型的关联规则挖掘通过将属性的定义域模糊化，从而达到更好的挖掘效果。该文提出一种基于云模型的检测组合发现方法，在大量的体检报告中进行关联挖掘，找出与癌症相关性最大的肿瘤指标。实验表明，挖掘出的前10个检测组合中，有80%符合目前的医学常识。

关键词: 数据挖掘, 关联规则, 云模型, 肿瘤标志物, 联合检测

CLC Number: 

• TP301.6