加权序列模式在临床异常行为检测中的应用

Application of weighted sequential pattern mining algorithms on detecting clinical abnormal behavior

目前临床异常行为检测的研究主要采用病症关联、费用控制和临床序列模式挖掘等方法。在临床序列模式挖掘方法中的临床医疗行为模式库的建立,只是简单的挖掘支持度满足阈值的频繁序列,忽略了出现频率低但在价格、用量上偏高的医疗行为。针对诊疗过程中出现频率少但属于盲目用贵药和过量用药等大处方医疗行为,文中提出改进的加权临床序列模式挖掘算法,挖掘正常临床医疗行为库的加权临床序列模式,为后续异常检测做准备。

Currently,the research on clinical abnormal behavior detection mainly uses the disease association,cost control and clinical sequence pattern mining method. Based on the clinical sequence pattern mining,the establishment of the clinical medical behavior pattern library,simply mines frequent sequences whose support meet the threshold,ignoring the medical behavior which with low frequency but high in price. For this problem,it puts forward the improved weighted clinical sequential patterns mining algorithm,mining the weighted clinical sequential patterns to prepare for the anomaly detection.