中文摘要：

针对挖掘高维临床数据的异常时存在着度量难及匹配性差的缺陷，提出一种医疗行为加权的最大频繁模式算法。通过对原始医疗数据集属性的整合，将医疗行为数据转换为结构化描述形式，解决数据的可匹配性；采用前缀树中节点中设置医疗权值的方法，解决医疗数据多样性带来的度量难问题；综合多种影响因子计算出医疗服务值，作为异常点评估依据。实验结果表明，该算法在给定合理的医疗权值情况下，能够有效的挖掘出临床数据中的异常现象。

英文摘要：

Considering the defects of difficult measurement and poor matching performance when mining from the high-dimensional clinical data, a algorithm of maximal frequent patterns confined by medical weight is proposed. To enhance the compatibility of data, structural description of the medical practices through the integration of the properties of the original medical data is used. The weight is set to the node of the prefix-tree to solve the problem that the medical data is hard to measure Finally, a variety of impact factors are used to calculate medical services value and to estimate the point whether it is an outlier. The experiments show that the improved algorithm could effectively find the anomalies from clinical data when reasonable medical weight is given.